˙
˙
Aca
d
e
m
ic E
d
ito
r
:
S
ta
n
i
s
ł
a
w
D
r
oz
d
z
Receive
d
:
11 Feb
ru
a
r
y 2025
Revi
s
e
d
:
22
M
a
r
c
h
2025
Acce
p
te
d
:
23
M
a
r
c
h
2025
P
u
bli
sh
e
d
:
27
M
a
r
c
h
2025
Citatio
n
:
Cao, J.
;
He, G.
;
Jiao, Y. Too
S
e
ns
itive to Fail
:
T
h
e I
mp
act o
f
S
e
n
ti
m
e
n
t Co
nn
ecte
d
n
e
ss
o
n
S
tock
P
r
ice C
r
a
sh
Ri
s
k. Ent
r
o
p
y
2025, 27, 345.
h
tt
ps
:
//
d
oi.o
r
g
/10.3390/e27040345
Co
p
y
r
i
gh
t
:
©
2025 by t
h
e a
u
t
h
o
rs
.
Lice
ns
ee
M
D
P
I,
B
a
s
el,
S
w
itze
r
la
n
d
.
T
h
i
s
a
r
ticle i
s
a
n
o
p
e
n
acce
ss
a
r
ticle
d
i
s
t
r
ib
u
te
d
un
d
e
r
t
h
e te
rms
a
n
d
co
n
d
itio
ns
o
f
t
h
e C
r
eative Co
mm
o
ns
Att
r
ib
u
tio
n
(
CC
B
Y
)
lice
ns
e
(h
tt
ps
:
//c
r
eativeco
mm
o
ns
.o
r
g
/
lice
ns
e
s
/by/4.0/
)
.
A
r
ticle
Too Se
ns
itive to
F
ail
:
T
h
e I
mp
act of Se
n
ti
m
e
n
t Co
nn
ecte
dn
e
ss
o
n
Stock
P
r
ice C
r
a
s
h
Ri
s
k
Jie Cao
1,
*
, G
u
o
q
i
ng
He
2
a
nd
Ya
p
i
ng
Jiao
2
1
S
c
h
ool o
f
Eco
n
o
m
ic
s
a
n
d
M
a
n
a
g
e
m
e
n
t, C
h
a
n
g
sh
a U
n
ive
rs
ity o
f
S
cie
n
ce a
n
d
Tec
hn
olo
g
y,
C
h
a
n
g
sh
a 410076, C
h
i
n
a
2
S
c
h
ool o
f
M
at
h
e
m
atic
s
a
n
d
S
tati
s
tic
s
, C
h
a
n
g
sh
a U
n
ive
rs
ity o
f
S
cie
n
ce a
n
d
Tec
hn
olo
g
y,
C
h
a
n
g
sh
a 410076, C
h
i
n
a
;
g
u
o
q
i
n
g
h
e824
@
163.co
m
(
G.H.
)
;
jiaoy
p
_1116
@
163.co
m
(
Y.J.
)
*
Co
rr
e
sp
o
n
d
e
n
ce
:
caojie_
m
at
h
@
126.co
m
A
bs
t
rac
t: U
s
i
n
g
a
sam
p
le o
f
S&P
500
s
to
c
k
s
, t
h
i
s
p
a
p
e
r
ex
am
i
n
e
s
t
h
e i
n
ve
s
to
r
s
e
n
ti
m
e
n
t
s
p
illove
r
n
et
w
o
r
k
b
et
w
ee
n
fi
r
ms
an
d
ass
e
ss
e
s
h
o
w
t
h
e
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
i
n
t
h
e
n
et
w
o
r
k i
mp
act
s
s
tock
pr
ice c
r
a
sh
r
i
s
k.
W
e
d
e
m
o
ns
t
r
ate t
h
at fi
rms
w
it
h
h
i
g
h
e
r
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
a
r
e
m
o
r
e likely to
c
r
ash
as
t
h
ey
s
pr
e
a
d
m
o
r
e i
rr
a
tio
na
l
s
e
n
ti
m
e
n
t
s
ig
na
l
s
an
d
a
r
e
m
o
r
e
s
e
ns
itive to i
n
ve
s
to
r
b
e
ha
vio
r
s
. Not
ab
ly,
w
e fi
n
d
t
ha
t t
h
e e
ff
e
c
t o
f
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t o
n
cr
a
sh
r
i
s
k
m
ai
n
ly
s
te
ms
fr
o
m
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
a
m
o
n
g
fi
rms
r
at
h
e
r
t
h
a
n
fi
r
ms’
i
n
d
ivi
du
al
s
e
n
ti
m
e
n
t, e
s
p
e
c
ially
w
h
e
n
m
a
r
ket
s
e
n
ti
m
e
n
t i
s
s
urg
i
n
g
o
r
d
e
c
li
n-
i
n
g. T
h
e
s
e fi
n
d
i
n
g
s
r
e
ma
i
n
r
o
b
u
s
t
af
te
r
c
o
n
t
r
olli
n
g
f
o
r
ot
h
e
r
d
ete
r
m
i
nan
t
s
o
f
c
r
ash
r
i
s
k,
i
n
cl
ud
i
n
g
stoc
k
pr
ic
e
sy
n
c
hr
o
n
icity, acco
un
ti
n
g
co
n
s
e
r
vatis
m
, a
nd
i
n
t
e
rn
al co
rp
o
r
at
e
g
ov
e
r-
n
a
n
c
e
s
t
r
e
n
g
t
h
.
O
ur
r
e
su
lt
s
und
e
rs
co
r
e
t
h
e
i
mp
o
r
ta
n
c
e
o
f
s
e
n
ti
m
e
n
t co
nn
e
ct
e
dn
e
ss
a
m
o
n
g
fi
rms
a
n
d
pr
ovi
d
e val
u
able i
ns
i
g
h
t
s
f
o
r
r
i
s
k
m
a
n
a
g
e
m
e
n
t a
m
o
n
g
i
n
ve
s
to
rs
a
n
d
r
e
g
u
lato
r
y
a
u
t
h
o
r
itie
s
i
n
volve
d
i
n
m
o
n
ito
r
i
n
g
r
i
s
k.
Key
w
o
r
d
s
:
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t
sp
illove
r
;
s
e
n
ti
m
e
n
t co
nn
ecte
d
n
e
ss
;
s
tock
pr
ice c
r
a
sh
r
i
s
k
1. I
n
t
r
o
du
ctio
n
I
n
ve
s
to
r
s
e
n
ti
m
e
n
t i
s
w
i
d
ely
r
e
c
og
n
ize
d
as
a
c
r
iti
ca
l
fac
to
r
b
e
h
i
n
d
ass
et
pr
i
c
e fl
u
c-
t
u
a
tio
ns
[
1–4
]
. Re
c
e
n
t fi
nanc
i
a
l
u
nc
e
r
t
a
i
n
tie
s
an
d
e
c
o
n
o
m
i
c
v
u
l
n
e
r
ab
ilitie
s
ha
ve le
d
to
i
nc
r
e
as
i
n
g
s
e
ns
itivity
an
d
w
i
d
e
s
pr
e
a
d
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t
an
d
p
an
i
c
d
i
ss
e
m
i
na
tio
n
,
r
e
-
s
u
lti
n
g i
n
s
eve
r
e e
c
o
n
o
m
i
c
c
o
ns
e
q
u
e
nc
e
s
[
5,6
]
. Fo
r
ex
am
p
le, t
h
e
fa
il
ur
e o
f
S
ili
c
o
n
V
a
lley
Ban
k o
n
Ma
r
ch
10, 2023,
has
r
a
i
s
e
d
s
ig
n
ifi
can
t
c
o
nc
e
r
ns
ab
o
u
t
b
r
o
a
d
e
r
ban
ki
n
g
s
t
ab
ility.
Co
ns
eq
u
e
n
tly, i
n
ve
s
to
rs
i
n
itiate
d
a
s
ell
-
o
ff
o
f
r
elate
d
s
tock
s
, ca
us
i
n
g
sh
a
rp
d
ecli
n
e
s
i
n
ove
r
20 ba
n
ki
n
g
e
q
u
iti
e
s
by
M
a
r
c
h
13, 2023. T
h
e
colla
ps
e
o
f
S
ilico
n
Vall
e
y
B
a
n
k a
nd
su
b
s
e
q
u
e
n
t
m
a
r
k
e
t co
n
c
e
rn
s
h
av
e
co
n
ti
nu
e
d
to
dr
a
w
att
e
n
tio
n
fr
o
m
i
n
v
e
sto
r
s a
nd
r
e
g
u
lato
r
s
r
e
g
a
rd
i
n
g
t
h
e t
r
ansm
i
ss
io
n
o
f
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t. A
n
ot
h
e
r
ex
am
p
le
d
e
m
o
ns
t
r
a
ti
n
g t
h
e i
n
te
r
c
o
n-
n
e
c
te
d
na
t
ur
e o
f
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t i
s
t
h
e
p
an
i
c
s
elli
n
g i
n
du
c
e
d
b
y
r
e
c
e
n
t geo
p
oliti
ca
l
c
o
n
fli
c
t
s
[
7–9
]
. T
h
e
s
e
cas
e
s
vivi
d
ly ill
u
s
t
r
a
te t
ha
t fi
nanc
i
a
l
ma
r
ket
s
a
r
e
s
u
sc
e
p
ti
b
le to
s
e
n
ti
m
e
n
t
an
d
it
s
c
o
nn
e
c
te
d
n
e
ss
. Colle
c
tive
p
an
i
c
am
o
n
g i
n
ve
s
to
r
s
,
dr
ive
n
b
y
ma
r
ket
i
nf
o
r
ma
tio
n
,
h
e
rd
b
e
ha
vio
r
,
an
d
s
e
c
to
r
-s
p
e
c
ifi
c
e
m
otio
na
l t
r
ansm
i
ss
io
n
,
ca
u
s
e
s
i
ns
t
ab
il
-
ity i
n
fi
nanc
i
a
l
ma
r
ket
s
[
10,11
]
. T
h
e
r
e
f
o
r
e,
s
t
ud
yi
n
g t
h
e i
m
p
ac
t o
f
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t
co
nn
ecte
d
n
e
ss
o
n
s
tock
pr
ice
s
i
s
c
ru
cial.
T
h
is
p
a
p
e
r
co
n
si
d
e
r
s t
h
e
co
nn
e
ct
e
dn
e
ss o
f
i
n
v
e
sto
r
s
e
n
ti
m
e
n
t b
e
t
w
ee
n
fi
rm
s a
nd
e
val
u-
at
e
s t
h
e
i
mp
act o
f
s
e
n
ti
m
e
n
t co
nn
e
ct
e
dn
e
ss o
n
fi
rm
c
r
as
h
r
is
k
.
B
a
k
e
r
a
nd
Wur
g
l
e
r
(
2007
)
[
6
]
h
ave co
n
fi
rm
e
d
t
h
at
s
tock
p
r
ice c
r
a
sh
r
i
s
k
m
ay be
d
r
ive
n
by i
n
ve
s
to
r
s
e
n
ti
m
e
n
t,
w
h
ic
h
i
s
Ent
r
o
p
y
2025, 27, 345
h
tt
ps
:
//
d
oi.o
r
g
/10.3390/e27040345
Ent
r
o
p
y
2025, 27, 345
2 o
f
26
o
f
te
n
r
e
g
a
r
d
e
d
a
s
f
acto
rs
beyo
n
d
t
h
e co
mp
a
n
y
fun
d
a
m
e
n
tal
s
(
e.
g
., li
qu
i
d
ity
[
12
]
, book
-
to
-
m
a
r
k
e
t
r
atio
[
13
])
.
W
e
st
(
1988
)
[
14
]
a
nd
Sh
ill
e
r
(
1990
)
[
15
]
a
r
g
u
e
t
h
at t
h
e
i
rr
atio
n
al
e
m
otio
n
s
o
f
i
n
v
e
sto
r
s, s
u
c
h
as
“
e
n
t
hu
sias
m
”
o
r
“
p
a
n
ic
”
,
h
av
e
t
h
e
p
ot
e
n
tial to t
r
i
gg
e
r
fl
u
ct
u
atio
n
s a
nd
b
u
bble
s
i
n
a
ss
et
pr
ice
s
. I
n
li
g
h
t o
f
t
h
i
s
,
s
o
m
e
s
t
u
d
ie
s
h
ave
sh
o
w
n
t
h
e
r
elatio
nsh
i
p
bet
w
ee
n
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t
an
d
c
r
ash
r
i
s
k,
s
u
ch
as
[
16–19
]
.
Wh
ile t
h
e
s
e vie
wp
oi
n
t
s
can
c
o
n
fi
r
m
t
h
e
r
elatio
nsh
i
p
bet
w
ee
n
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t a
n
d
s
tock
pr
ice colla
ps
e
s
, t
h
ey o
f
te
n
ove
r
look
t
h
e co
nn
ecte
d
n
e
ss
o
f
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t
s
,
w
h
ic
h
ca
n
s
o
m
eti
m
e
s
h
i
n
d
e
r
t
h
e acc
u
r
ate i
d
e
n-
tifi
c
atio
n
o
f
t
h
e i
m
p
a
c
t o
f
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t. T
h
e
c
o
nn
e
c
te
d
n
e
ss
p
h
e
n
o
m
e
n
o
n
i
n
d
i
c
ate
s
t
h
at
s
e
n
ti
m
e
n
t
sh
ock ca
n
s
p
r
ea
d
fr
o
m
o
n
e fi
rm
to a
n
ot
h
e
r
t
hr
o
ug
h
fi
n
a
n
cial li
n
ka
g
e
s
[
20
]
.
I
nd
ee
d
, i
n
v
e
sto
r
s
e
n
ti
m
e
n
t
h
as b
ee
n
s
h
o
wn
to b
e
s
e
n
sitiv
e
to
e
x
t
e
rn
al i
n
fl
u
e
n
c
e
s a
nd
e
x
h
ibit
s
p
illove
r
ac
r
o
ss
fi
rms
i
n
fi
n
a
n
cial
m
a
r
ket
s
[
6,20,21
]
. T
h
i
s
m
ea
ns
t
h
at t
h
e
s
e
n
ti
m
e
n
t o
f
o
n
e
fi
rm
i
s
n
ot
s
olely i
n
fl
u
e
n
ce
d
by it
s
o
w
n
s
e
n
ti
m
e
n
t, b
u
t it ca
n
al
s
o be
s
i
g
n
ifica
n
tly i
mp
acte
d
by t
h
e
s
e
n
ti
m
e
n
t
s
o
f
ot
h
e
r
fi
rms
.
T
h
i
s
p
a
p
e
r
u
s
e
s
t
h
e
n
o
n
li
n
e
a
r
G
r
an
ge
r
ca
u
sa
lity
m
et
h
o
d
[
22
]
to
s
t
ud
y t
h
e
s
e
n
ti
-
m
e
n
t
s
p
illove
r
n
et
w
o
r
k
am
o
n
g
S&P
500
s
to
c
k
s
. T
h
e
s
e
n
ti
m
e
n
t
s
p
illove
r
n
et
w
o
r
k
d
i
f-
f
e
r
s
f
r
o
m
pr
evio
u
s
fi
nanc
i
a
l li
n
k
a
ge
s
bas
e
d
o
n
b
u
s
i
n
e
ss
c
o
nn
e
c
tio
ns
[
23–25
]
o
r
pr
i
c
e
c
o
-
m
ove
m
e
n
t
s
[
22,26,27
]
. Fi
rs
t, t
h
e fi
rm
s
e
n
ti
m
e
n
t co
nn
ecte
d
n
e
ss
i
s
d
r
ive
n
by t
h
e e
x
o
g
e
n
o
us
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t
f
acto
r
, o
f
te
n
r
e
g
a
r
d
e
d
a
s
i
rr
atio
n
al a
n
d
w
eakly
r
elate
d
to co
mp
a
n
y
fun-
d
a
m
e
n
tal
s
[
6
]
.
S
eco
n
d
, t
h
e
s
e
n
ti
m
e
n
t co
nn
ecte
d
n
e
ss
o
f
a fi
rm
ca
nn
ot be e
n
ti
r
ely att
r
ib
u
te
d
to it
s
fi
n
a
n
cial c
h
a
r
act
e
r
i
s
tic
s
.
F
o
r
i
ns
ta
n
c
e
, i
n
t
h
e
co
ns
t
ru
ct
e
d
s
e
n
ti
m
e
n
t
sp
illov
e
r
n
e
t
w
o
r
k,
Am
a
z
o
n
i
s
li
n
ke
d
to
m
a
n
y ot
h
e
r
fi
r
ms
,
w
h
e
r
ea
s
J
P
M
o
rg
a
n
Ch
a
se a
n
d
G
o
ldm
a
n
S
a
chs
h
ave
mu
c
h
le
ss
s
e
n
ti
m
e
n
t co
nn
ecte
dn
e
ss
d
e
sp
ite bei
n
g
r
a
n
ke
d
a
s
g
lobal
s
y
s
te
m
ically i
mp
o
r
ta
n
t
fi
n
a
n
cial i
ns
tit
u
tio
ns
. Take
n
to
g
et
h
e
r
, o
ur
fi
n
d
i
n
g
s
su
gg
e
s
t t
h
at t
h
e i
n
ve
s
to
r
s
e
n
ti
m
e
n
t o
f
a
fi
rm
ca
n
r
eceive
fr
o
m
o
r
s
p
r
ea
d
to ot
h
e
r
fi
rms
, eve
n
w
it
h
o
u
t
d
i
r
ectly a
ff
ecti
n
g
o
w
n
e
rsh
i
p
o
r
a
ss
et flo
w
s
.
W
e
m
e
as
ur
e t
h
e
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
o
f
a
fi
r
m
bas
e
d
o
n
d
i
ff
e
r
e
n
t
n
et
w
o
r
k
c
e
n-
t
r
a
lity
m
e
as
ur
e
m
e
n
t
s
. A
cc
o
rd
i
n
g to El
-
K
ha
ti
b
et
a
l.
(
2015
)
[
24
]
, t
h
e
c
o
nn
e
c
te
d
n
e
ss
o
f
an
i
n
d
ivi
du
a
l i
n
a
n
et
w
o
r
k
ass
u
m
e
s
n
o
d
e
s
(
1
)
a
r
e li
n
ke
d
to
man
y ot
h
e
r
fi
r
ms
;
(
2
)
a
r
e
c
lo
s
e
to ot
h
e
r
i
n
d
ivi
d
u
al
s
;
(
3
)
a
r
e locate
d
o
n
t
h
e
p
at
h
co
nn
ecti
n
g
a
n
y ot
h
e
r
p
ai
rs
o
f
i
n
d
ivi
d
u
al
s
;
an
d
(
4
)
a
r
e li
n
ke
d
to ot
h
e
r
h
ig
h
ly li
n
ke
d
i
n
d
ivi
du
a
l
s
.
M
otiv
a
te
d
b
y t
h
i
s
,
man
y
s
t
ud
ie
s
h
av
e
s
h
o
wn
t
h
e
st
ru
ct
ur
e
a
nd
i
mp
act o
f
fi
n
a
n
cial
n
e
t
w
o
r
k
s
u
si
n
g
d
e
g
r
ee
c
e
n
t
r
ality
[
28–30
]
,
clo
s
e
n
e
ss
ce
n
t
r
ality
[
24,31
]
, bet
w
ee
nn
e
ss
ce
n
t
r
ality
[
32
]
, ei
g
e
n
vecto
r
ce
n
t
r
ality
[
33,34
]
, a
n
d
s
o o
n
. T
h
e
s
e
i
nd
e
x
e
s
m
e
a
sur
e
t
h
e
c
e
n
t
r
ality o
r
e
x
t
e
n
t o
f
a fi
rm’s
co
nn
e
ct
e
dn
e
ss
w
it
h
ot
h
e
rs
a
n
d
t
h
e
p
o
ss
ible i
nf
o
rm
atio
n
a
n
d
r
e
s
o
ur
ce e
x
c
h
a
n
g
e
[
35
]
. Ho
w
eve
r
, t
h
e
s
e ce
n
t
r
alitie
s
o
n
ly
co
ns
i
d
e
r
o
n
e ki
n
d
o
f
n
et
w
o
r
k li
n
ka
g
e a
n
d
i
n
fl
u
e
n
ce, ca
us
i
n
g
bia
s
a
n
d
violati
n
g
m
ea
sur
e
s
o
f
n
e
t
w
o
r
k
r
e
latio
nsh
i
ps
[
36
]
;
s
o
m
e
ti
m
e
s
, t
h
e
s
e
c
e
n
t
r
aliti
e
s
a
r
e
i
n
t
r
i
ns
ically
r
e
l
e
va
n
t
[
37,38
]
.
To
e
li
m
i
n
at
e
s
u
c
h
bias
e
s a
nd
m
e
as
ur
e
d
i
ff
e
r
e
n
t i
n
fl
u
e
n
c
e
s o
f
a
n
o
d
e
,
w
e
a
gg
r
e
g
at
e
d
i
ff
e
r
e
n
t
n
et
w
o
r
k ce
n
t
r
ality i
n
d
e
x
e
s
ba
s
e
d
o
n
t
h
e e
n
t
r
o
p
y
w
ei
g
h
t
m
et
h
o
d
[
39
]
a
n
d
pr
o
p
o
s
e a
n
ovel
fi
r
m
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
i
n
d
ex,
w
h
i
ch
i
s
r
e
f
e
rr
e
d
to
as
Fi
r
m
S
e
n
tix, to
m
e
as
ur
e t
h
e
co
nn
ectio
n
a
n
d
sp
illove
r
o
f
eac
h
fi
rm
.
W
e e
x
a
m
i
n
e t
h
e
r
elatio
nsh
i
p
bet
w
ee
n
s
e
n
ti
m
e
n
t co
nn
ecte
d
n
e
ss
a
n
d
s
tock
pr
ice c
r
a
sh
r
i
s
k i
n
S&P
500
s
to
c
k
s
w
it
h
s
eve
r
a
l
r
eg
r
e
ss
io
ns
.
W
e fi
n
d
t
ha
t
a
fi
r
m’s
s
e
n
ti
m
e
n
t
c
o
n-
n
e
c
te
d
n
e
ss
i
s
s
ig
n
ifi
can
tly
an
d
p
o
s
itively
r
el
a
te
d
to
c
r
ash
r
i
s
k,
sh
o
w
i
n
g t
ha
t
a
fi
r
m
w
it
h
h
i
g
h
e
r
co
nn
ecte
d
n
e
ss
i
n
t
h
e
s
e
n
ti
m
e
n
t
n
et
w
o
r
k i
s
m
o
r
e v
u
l
n
e
r
able to c
r
a
sh
e
s
. T
h
i
s
r
e
s
u
lt
d
e
m
o
ns
t
r
a
te
s
t
ha
t
s
to
c
k
pr
i
c
e i
s
n
ot o
n
ly
r
el
a
te
d
to
a
fi
r
m’s
i
n
d
ivi
du
a
l
s
e
n
ti
m
e
n
t,
b
u
t
a
l
s
o
s
u
sc
e
p
ti
b
le to
s
e
n
ti
m
e
n
t i
n
te
r
ac
tio
ns
w
it
h
it
s
n
eig
hb
o
r
s
. T
h
e
r
el
a
tio
nsh
i
p
b
et
w
ee
n
s
e
n
ti
m
e
n
t co
nn
e
ct
e
dn
e
ss a
nd
c
r
as
h
r
is
k
ca
n
b
e
e
x
p
lai
n
e
d
i
n
t
w
o
d
i
m
e
n
sio
n
s.
F
i
r
st, a
h
i
g
h
ly
i
n
te
r
co
nn
ecte
d
fi
rm
m
ay be
m
o
r
e i
n
fl
u
e
n
tial i
n
spr
ea
d
i
n
g
s
e
n
ti
m
e
n
t
sh
ock
s
a
s
it ca
n
m
o
r
e
e
f
ficie
n
tly
r
eac
h
ot
h
e
r
m
e
m
be
rs
o
f
t
h
e
n
et
w
o
r
k.
S
eco
n
d
, a
h
i
g
h
ly co
nn
ecte
d
m
e
m
be
r
m
ay
b
e
m
o
r
e
e
asily i
n
fl
u
e
n
c
e
d
by ot
h
e
r
s, as its
n
e
t
w
o
r
k
p
ositio
n
pr
ovi
d
e
s
m
o
r
e
o
pp
o
r
t
un
iti
e
s to
Ent
r
o
p
y
2025, 27, 345
3 o
f
26
co
n
tact
p
ee
r
fi
rms
, allo
w
i
n
g
i
m
itatio
n
o
f
be
h
avio
rs
a
nd
f
o
s
te
r
i
n
g
i
rr
atio
n
al
d
y
n
a
m
ic
s
. T
h
i
s
fi
n
d
i
n
g
r
e
m
ai
ns
r
ob
us
t
w
h
e
n
co
ns
i
d
e
r
i
n
g
ot
h
e
r
d
ete
rm
i
n
a
n
t
s
o
f
c
r
a
sh
r
i
s
k, i.e.,
s
tock
pr
ice
s
y
nch
r
o
n
i
c
ity,
acc
o
u
n
ti
n
g
c
o
ns
e
r
v
a
ti
sm
, ext
r
e
m
e
ma
r
ket eve
n
t
s
,
an
d
i
n
te
r
na
l
c
o
rp
o
r
a
te
gove
r
nanc
e
s
t
r
e
n
gt
h
. To
a
ddr
e
ss
p
ote
n
ti
a
l e
n
d
oge
n
eity i
ss
u
e
s
, t
w
o
-s
t
a
ge le
as
t
sq
u
a
r
e
s
e
s
ti
m
atio
n
a
n
d
pr
o
p
e
ns
ity
s
co
r
e
m
atc
h
i
n
g
a
r
e e
mp
loye
d
. A
f
te
r
co
ns
i
d
e
r
i
n
g
t
h
e i
mp
act o
f
p
ee
r
e
ff
e
c
t, i.e., t
h
e
p
ee
r
c
o
m
p
an
ie
s’
c
r
ash
,
an
d
h
ig
h-f
r
e
q
u
e
nc
y t
r
a
d
i
n
g, t
h
e
r
el
a
tio
nsh
i
p
bet
w
ee
n
s
e
n
ti
m
e
n
t co
nn
ecte
d
n
e
ss
a
n
d
c
r
a
sh
r
i
s
k
r
e
m
ai
ns
r
ob
us
t.
To
d
e
m
o
ns
t
r
ate t
h
e a
d
va
n
ta
g
e o
f
c
o
ns
i
d
e
r
i
n
g
t
h
e
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
b
et
w
ee
n
fi
r
ms
,
w
e
c
o
m
p
a
r
e t
h
e
s
e
ns
itivity o
f
fi
r
ms’
i
n
d
ivi
du
a
l
s
e
n
ti
m
e
n
t
an
d
t
h
e
s
e
n
ti
m
e
n
t
c
o
n-
n
e
c
te
d
n
e
ss
to
c
r
a
sh
r
i
s
k
u
n
d
e
r
d
i
ff
e
r
e
n
t
m
a
r
ket
c
o
n
d
itio
ns
. Co
ns
i
d
e
r
i
n
g
t
h
at
s
p
illove
r
i
s
m
o
r
e
li
k
e
ly to occ
ur
wh
e
n
s
e
n
ti
m
e
n
t b
e
co
m
e
s
e
x
t
r
e
m
e
[
40,41
]
,
w
e
d
ivi
d
e
t
h
e
wh
ol
e
sa
mp
l
e
i
n
to
p
e
r
io
d
s
o
f
h
eig
h
te
n
e
d
an
d
d
i
m
i
n
i
sh
e
d
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t
bas
e
d
o
n
t
h
e
m
e
d
i
an
o
f
t
h
e fi
r
m’s
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t. O
ur
r
e
s
u
lt
sh
o
w
s
t
ha
t
b
ot
h
t
h
e fi
r
m’s
s
e
n
ti
m
e
n
t
an
d
t
h
e
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
a
r
e
s
ig
n
ifi
can
tly
an
d
p
o
s
itively
r
el
a
te
d
to
c
r
ash
r
i
s
k
ac
r
o
ss
t
h
e
fu
ll
s
a
mp
l
e
, i
nd
icati
n
g
t
h
at
s
e
n
ti
m
e
n
t a
nd
t
h
e
co
nn
e
ct
e
dn
e
ss
m
ay i
n
c
r
e
a
s
e
t
h
e
fi
rm’s
c
r
a
sh
r
i
s
k. Ho
w
eve
r
, a fi
rm’s
i
n
d
ivi
d
u
al
s
e
n
ti
m
e
n
t i
s
n
ot a
s
i
g
n
ifica
n
t
d
r
ive
r
o
f
c
r
a
sh
r
i
s
k
w
h
e
n
m
a
r
ket
s
e
n
ti
m
e
n
t i
s
at a
h
i
g
h
level o
r
lo
w
level. I
n
co
n
t
r
a
s
t,
s
e
n
ti
m
e
n
t
sp
illove
r
r
e
m
ai
ns
a
s
i
g
n
ifica
n
t i
n
fl
u
e
n
ce o
n
c
r
a
sh
r
i
s
k. T
h
i
s
ca
n
be e
x
p
lai
n
e
d
by t
h
e
f
act t
h
at
dur
i
n
g
p
e
r
io
ds
o
f
e
x
t
r
e
m
e
s
e
n
ti
m
e
n
t, i
n
v
e
sto
r
s
e
n
ti
m
e
n
t ca
n
b
e
fu
lly
r
e
fl
e
ct
e
d
by t
h
e
m
a
r
k
e
t
[
6
]
, so t
h
e
i
mp
act
o
f
s
e
n
ti
m
e
n
t fl
u
c
t
u
a
tio
ns
o
n
s
to
c
k
pr
i
c
e vol
a
tility i
s
r
el
a
tively
sma
ll.
M
o
r
eove
r
,
w
h
e
n
m
a
r
ket
s
e
n
ti
m
e
n
t i
s
h
ei
g
h
te
n
i
n
g
a
n
d
d
i
m
i
n
i
sh
i
n
g
, i
n
ve
s
to
rs
g
e
n
e
r
ally
h
ol
d
o
p
ti
m
i
s
ti
c
o
r
p
e
ss
i
m
i
s
tic vie
w
s
,
m
aki
n
g
t
h
e
m
m
o
r
e likely to i
g
n
o
r
e
n
e
g
ative i
nf
o
rm
atio
n
o
r
r
i
s
k
f
acto
rs
,
t
h
e
r
e
b
y l
ac
ki
n
g
s
e
ns
itivity to
p
ote
n
ti
a
l
c
r
ash
r
i
s
k
s
. Ho
w
eve
r
,
w
h
e
n
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t
spr
ea
d
s
a
n
d
pr
o
p
a
g
ate
s
t
hr
o
ug
h
co
mm
u
n
icatio
n
o
r
m
e
d
ia cove
r
a
g
e, ot
h
e
r
i
n
ve
s
to
rs
m
ay
al
s
o be a
ff
ecte
d
, t
h
e
r
eby e
x
ace
r
bati
n
g
fl
u
ct
u
atio
ns
a
n
d
c
h
a
n
g
e
s
i
n
m
a
r
ket
s
e
n
ti
m
e
n
t. T
h
i
s
i
n
d
icate
s
t
h
at i
n
ve
s
to
r
s
e
n
ti
m
e
n
t co
nn
ecte
d
n
e
ss
h
a
s
s
t
r
o
n
g
e
r
e
x
p
la
n
ato
r
y
p
o
w
e
r
f
o
r
c
r
a
sh
r
i
s
k t
h
a
n
a fi
rm’s
i
n
d
ivi
d
u
al
s
e
n
ti
m
e
n
t.
W
e
f
ur
t
h
e
r
sh
o
w
p
ote
n
ti
a
l
chann
el
s
t
ha
t
aff
e
c
t t
h
e
r
el
a
tio
nsh
i
p
b
et
w
ee
n
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
an
d
fi
r
m
c
r
ash
r
i
s
k.
M
o
d
e
r
a
ti
n
g e
ff
e
c
t
ana
ly
s
i
s
s
u
gge
s
t
s
t
ha
t
s
to
c
k
pr
i
c
e
sy
n
c
hr
o
n
icity a
mp
lifi
e
s t
h
e
e
ff
e
ct o
f
n
e
t
w
o
r
k
co
nn
e
ctio
n
s o
n
c
r
as
h
r
is
k
, b
u
t acco
un
ti
n
g
co
n-
s
e
r
vati
sm
le
ss
e
ns
it. T
h
i
s
i
s
beca
us
e
s
tock
pr
ice
s
y
n
c
hr
o
n
icity i
n
c
r
ea
s
e
s
i
nf
o
rm
atio
n
a
s
y
m-
m
et
r
y
an
d
r
a
i
s
e
s
i
n
ve
s
to
r
s’
s
e
ns
itivity to
ba
d
n
e
w
s
. Ho
w
eve
r
,
acc
o
u
n
ti
n
g
c
o
ns
e
r
v
a
ti
sm
r
e
s
t
r
ict
s
e
x
e
c
u
tiv
e
s’
o
pp
o
r
t
un
iti
e
s
to co
n
c
e
al ba
d
n
e
ws
, i
n
c
r
e
a
s
e
s
i
nf
o
rm
atio
n
t
r
a
nsp
a
r
e
n
cy,
a
n
d
r
e
d
u
ce
s
i
n
ve
s
to
r
s
e
ns
itivity
[
42–44
]
.
F
ur
t
h
e
rm
o
r
e, t
hr
o
u
g
h
h
ete
r
o
g
e
n
eity a
n
aly
s
i
s
,
w
e
sh
o
w
t
ha
t ext
r
e
m
e
r
i
s
k eve
n
t
s
an
d
p
oo
r
c
o
m
p
an
y gove
r
nanc
e
can
i
m
pr
ove t
h
e e
ff
e
c
t o
f
s
e
n
ti
m
e
n
t co
nn
ecte
d
n
e
ss
o
n
c
r
a
sh
r
i
s
k.
T
h
e
co
n
t
r
ib
u
tio
n
o
f
t
h
is
p
a
p
e
r
ca
n
b
e
co
n
cl
ud
e
d
fr
o
m
t
hr
ee
as
p
e
cts.
F
i
r
st, t
h
e
s
e
n
ti
m
e
n
t
co
nn
ecte
d
n
e
ss
a
m
o
n
g
fi
rms
i
s
s
t
ud
ie
d
, a
n
d
a co
rr
e
sp
o
n
d
i
n
g
s
e
n
ti
m
e
n
t
sp
illove
r
n
et
w
o
r
k i
s
c
o
ns
t
ru
c
te
d
.
Wh
ile
man
y
pr
io
r
s
t
ud
ie
s
f
o
c
u
s
o
n
t
h
e e
c
o
n
o
m
i
c
i
m
p
ac
t
s
o
f
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t
s
(s
ee, e.
g
., K
um
a
r
a
n
d
Lee
(
2006
)
[
45
]
, De et al.
(
1990
)
[
46
]
, a
n
d
Ha
n
(
2008
)
[
47
])
,
t
h
e
y ov
e
r
loo
k
t
h
e
i
n
t
e
rp
lay o
f
i
nd
ivi
du
al s
e
n
ti
m
e
n
ts. T
h
is
p
a
p
e
r
i
n
v
e
sti
g
at
e
s t
h
e
s
e
n
ti
m
e
n
t
i
n
t
e
r
actio
n
s b
e
t
w
ee
n
fi
rm
s
u
si
n
g
t
h
e
n
e
t
w
o
r
k
st
ru
ct
ur
e
a
nd
m
e
as
ur
e
s t
h
e
i
n
v
e
sto
r
s
e
n
ti
m
e
n
t
co
nn
e
ct
e
dn
e
ss
w
it
h
a
n
ov
e
l a
gg
r
e
g
at
e
d
c
e
n
t
r
ality i
nd
e
x
.
O
ur
s
e
n
ti
m
e
n
t
n
e
t
w
o
r
k
s
h
o
w
s t
h
at
t
h
e
co
n
ta
g
io
n
o
f
i
n
v
e
sto
r
s
e
n
ti
m
e
n
t is
n
ot
m
e
r
e
ly
dr
iv
e
n
by t
h
e
fi
rm
fi
n
a
n
cial c
h
a
r
act
e
r
istics
su
c
h
a
s
s
ize,
w
h
ic
h
m
ea
ns
t
h
at t
h
e
s
e
n
ti
m
e
n
t
n
et
w
o
r
k ca
n
ca
p
t
ur
e t
h
e i
n
te
r
actio
n
bet
w
ee
n
i
n
ve
s
to
r
s
. I
m
p
o
r
t
an
tly, t
h
e
s
t
ru
c
t
ur
e o
f
t
h
e
s
e
n
ti
m
e
n
t
n
et
w
o
r
k
a
l
s
o
r
efle
c
t
s
t
h
e
ma
r
ket
d
y
n
a
m
ic
s
, a
s
t
h
e
n
et
w
o
r
k i
s
m
o
r
e
h
ete
r
o
g
e
n
eo
us
i
n
c
r
i
s
i
s
yea
rs
. T
h
e
s
e fi
n
d
i
n
g
s
sh
o
w
t
h
at
t
h
e
s
e
n
ti
m
e
n
t
n
et
w
o
r
k i
s
us
e
fu
l a
n
d
i
mp
o
r
ta
n
t i
n
d
e
s
c
r
ibi
n
g
t
h
e fi
n
a
n
cial
m
a
r
ket.
S
e
c
o
n
d
, t
h
i
s
p
a
p
e
r
ex
p
an
d
s
t
h
e i
n
ve
s
tig
a
tio
n
o
f
fi
r
m
c
r
ash
r
i
s
k
b
y
c
o
ns
i
d
e
r
i
n
g t
h
e
e
ff
ect o
f
s
e
n
ti
m
e
n
t co
nn
ecte
d
n
e
ss
a
m
o
n
g
fi
rms
.
P
r
io
r
r
e
s
ea
r
c
h
h
a
s
pr
e
d
o
m
i
n
a
n
tly co
n
ce
n-
Ent
r
o
p
y
2025, 27, 345
4 o
f
26
t
r
ate
d
o
n
a
n
alyzi
n
g
fun
d
a
m
e
n
tal
f
acto
rs
su
c
h
a
s
m
a
n
a
g
e
r
ial c
h
a
r
acte
r
i
s
tic
s
a
n
d
co
mp
a
n
y
fun
d
a
m
e
n
tal
s
(s
ee,
[
48–50
]
, etc.
)
. Ho
w
eve
r
, t
h
e
pr
e
s
e
n
ce o
f
fi
n
a
n
cial
m
a
r
ket lo
w
-
e
f
ficie
n
cy
a
n
d
s
to
c
k
p
r
i
c
e
s
y
nchr
o
n
i
c
ity i
n
d
i
c
ate
s
t
h
at t
h
e
r
i
s
k o
f
s
to
c
k
m
a
r
ket
cr
a
sh
e
s
i
s
fr
e
q
u
e
n
tly
i
n
fl
u
e
n
c
e
d
by
f
acto
r
s b
e
yo
nd
co
mp
a
n
y
fund
a
m
e
n
tals
[
14,51
]
. T
h
is
p
oi
n
t
h
as b
ee
n
vali
d
at
e
d
b
y
n
u
m
e
r
o
u
s
r
e
a
l
-
li
f
e
c
r
i
s
i
s
eve
n
t
s
,
f
o
r
ex
am
p
le, t
h
e
ban
k
p
an
i
c
f
ollo
w
i
n
g t
h
e
fa
il
ur
e
s
o
f
Silic
o
n
V
a
lle
y
B
a
n
k
an
d
Si
gn
a
t
ur
e
B
a
n
k. De
s
p
ite t
h
e i
m
p
ac
t
s
o
f
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t
b
ei
n
g
sh
o
w
n
i
n
s
o
m
e
s
t
u
d
ie
s
, t
h
e
sp
illove
r
a
n
d
n
et
w
o
r
k
s
t
ru
ct
ur
e o
f
s
e
n
ti
m
e
n
t a
r
e
n
e
g
lecte
d
. I
n
t
h
is
p
a
p
e
r
,
w
e
s
h
o
w
t
h
at t
h
e
co
nn
e
ct
e
dn
e
ss o
f
fi
rm
s
e
n
ti
m
e
n
t,
wh
ic
h
is b
e
yo
nd
fund
a
m
e
n
tal
f
acto
rs
,
m
ay ca
us
e a fi
rm
c
r
a
sh
.
O
ur
s
t
u
d
y e
nr
ic
h
e
s
t
h
e
r
ea
s
o
n
f
o
r
fi
rm
c
r
a
sh
r
i
s
k.
Fi
n
ally, t
h
i
s
p
a
p
e
r
d
e
m
o
ns
t
r
ate
s
t
h
at a fi
rm’s
s
e
n
ti
m
e
n
t co
nn
ecte
d
n
e
ss
h
ol
d
s
g
r
eate
r
ex
p
l
ana
to
r
y
p
o
w
e
r
r
eg
a
rd
i
n
g
c
r
ash
r
i
s
k t
han
it
s
i
n
d
ivi
du
a
l
s
e
n
ti
m
e
n
t. O
ur
s
t
ud
y i
n
te
-
g
r
a
te
s
i
n
ve
s
to
r
i
n
te
r
ac
tio
n
n
et
w
o
r
k
s
i
n
to t
h
e
r
e
s
e
a
r
ch
o
n
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t,
r
eve
a
li
n
g
t
h
at t
h
e
s
e
n
et
w
o
r
k
s
s
i
g
n
ifi
c
a
n
tly i
n
fl
u
e
nc
e i
n
ve
s
to
r
b
e
h
avio
r
.
S
o
c
ial
n
et
w
o
r
k
s
a
cc
ele
r
ate
i
nf
o
rm
atio
n
d
i
ss
e
m
i
n
atio
n
, e
nh
a
n
ci
n
g
it
s
i
mp
act. A
n
tici
p
ati
n
g
t
h
i
s
spr
ea
d
h
el
ps
us
bette
r
und
e
r
sta
nd
m
a
r
k
e
t
d
y
n
a
m
ics.
M
o
r
e
ov
e
r
,
d
i
ff
e
r
e
n
c
e
s i
n
h
o
w
i
n
v
e
sto
r
s i
n
t
e
rpr
e
t i
nf
o
rm
atio
n
w
it
h
i
n
s
ocial
n
et
w
o
r
k
s
t
r
i
gg
e
r
t
r
a
d
i
n
g
activitie
s
(
P
e
n
g
a
n
d
Z
h
a
n
g
, 2024
[
52
])
. U
s
i
n
g
s
ocial
n
et
w
o
r
k
ana
ly
s
i
s
,
w
e
can
s
t
ud
y i
n
ve
s
to
r
b
e
ha
vio
r
an
d
it
s
e
ff
e
c
t
s
o
n
t
r
a
d
i
n
g vol
u
m
e
an
d
m
a
r
ket li
qu
i
d
ity.
T
h
e
r
e
ma
i
n
i
n
g
s
e
c
tio
ns
o
f
t
h
i
s
p
a
p
e
r
a
r
e
s
t
ru
c
t
ur
e
d
as
f
ollo
w
s
.
S
e
c
tio
n
2
pr
e
s
e
n
t
s
d
e
fi
n
itio
n
s a
nd
m
e
t
h
o
d
s
p
e
r
ti
n
e
n
t to
e
mp
i
r
ical a
n
alysis.
S
e
ctio
n
3
d
isc
u
ss
e
s t
h
e
d
ata so
ur
c
e
an
d
f
u
n
d
am
e
n
t
a
l
c
o
n
d
itio
ns
.
S
e
c
tio
n
4
pr
e
s
e
n
t
s
t
h
e e
m
p
i
r
i
ca
l fi
n
d
i
n
g
s
. L
as
tly,
S
e
c
tio
n
5
summ
a
r
ize
s
t
h
e co
n
cl
us
io
ns
.
2.
M
et
h
o
d
olo
g
y
2.1. Sentiment S
p
ill
o
v
e
r
Netw
o
r
k
2.1.1.
S
e
n
ti
m
e
n
t
Sp
illove
r
Net
w
o
r
k Co
ns
t
ru
ctio
n
T
h
i
s
p
a
p
e
r
us
e
s
t
h
e t
urn
ove
r
r
ate a
s
t
h
e
m
ea
sur
e o
f
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t. T
h
e t
urn
ove
r
r
a
te
can
r
e
pr
e
s
e
n
t t
h
e
h
ete
r
oge
n
eity o
f
i
n
ve
s
to
r
’s
b
elie
fs
i
n
t
h
e
ma
r
ket
an
d
ca
p
t
ur
e t
h
e
liq
u
i
d
ity o
f
i
nd
ivi
du
al stoc
k
s. A
h
i
g
h
e
r
t
urn
ov
e
r
r
at
e
also o
f
t
e
n
m
e
a
n
s
m
o
r
e
h
ot
m
a
r
k
e
t a
nd
h
i
g
h
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t
[
53
]
. I
n
p
a
r
tic
u
la
r
,
w
h
e
n
i
rr
atio
n
ally o
p
ti
m
i
s
tic i
n
ve
s
to
r
s
e
n
ti
m
e
n
t
r
i
s
e
s
du
e to
sh
o
r
t
-s
elli
n
g
c
o
ns
t
r
a
i
n
t
s
,
r
a
tio
na
l i
n
ve
s
to
r
s
w
ill
ac
tively t
r
a
d
e to
pr
ofit
f
r
o
m
t
h
ei
r
d
ive
r
g
e
n
ce i
n
vie
w
s
[
6
]
.
W
e calc
u
late t
h
e t
urn
ove
r
r
ate a
s
t
h
e
r
atio o
f
r
e
p
o
r
te
d
sh
a
r
e
vol
um
e to ci
r
c
u
lati
n
g
sh
a
r
e
s
o
f
s
tock i o
n
d
ay t.
I
n
t
h
i
s
p
a
p
e
r
,
w
e
c
o
ns
t
ru
c
t
s
e
n
ti
m
e
n
t
s
p
illove
r
b
et
w
ee
n
fi
r
ms
u
s
i
n
g t
h
e
n
o
n
li
n-
e
a
r
G
r
an
ge
r
n
et
w
o
r
k
m
et
h
o
d
pr
o
p
o
s
e
d
i
n
[
22
]
. Fo
r
a
sam
p
le o
f
t
h
e t
ur
n
ove
r
o
f
S&P
500 co
mp
o
n
e
n
t
s
tock
s
ov
e
r
t
h
e
p
e
r
io
d
fr
o
m
Ja
nu
a
r
y 2006 to D
e
c
e
m
b
e
r
2021, t
h
e
pr
oc
e
dur
e
i
n
cl
u
d
e
s
t
h
e
f
ollo
w
i
n
g
s
te
ps
:
Fi
rs
t,
w
e filte
r
li
n
ea
r
ca
u
s
ality
u
s
i
n
g
t
h
e VAR
m
o
d
el, a
n
d
u
s
e VAR
r
e
s
i
du
al
s
to ca
rr
y
o
u
t t
h
e
n
o
n
li
n
e
a
r
G
r
an
ge
r
ca
u
sa
lity te
s
t
c
o
nc
r
etely
f
o
r
s
t
a
tio
na
r
y ti
m
e
s
e
r
ie
s
x
t
an
d
y
t
.
S
i
m
ila
r
to
[
54
]
,
w
e e
s
tabli
sh
t
h
e
f
ollo
w
i
n
g
li
n
ea
r
biva
r
iate VAR
:
x
t
=
a
10
−
b
10
y
t
+
b
11
x
t−1
+
b
12
y
t−1
+
ε
x
t
,
y
t
=
a
20
−
b
20
x
t
+
b
21
x
t−1
+
b
22
y
t−1
+
ε
y
t
,
(
1
)
w
h
e
r
e
x
t
an
d
y
t
a
r
e
s
t
a
tio
na
r
y ti
m
e
s
s
e
r
ie
s
an
d
t
h
e o
p
ti
ma
l l
a
g o
rd
e
r
o
f
t
h
e VAR(n)
ac-
co
r
d
i
n
g
to t
h
e Akaike i
nf
o
rm
atio
n
c
r
ite
r
io
n
. T
h
e
n
,
w
e obtai
n
e
d
t
h
e
r
e
s
i
d
u
al
s
e
qu
e
n
ce
s
ε
x
t
a
n
d
ε
y
t
.
S
eco
n
d
, a
B
D
S
te
s
t i
s
p
e
rf
o
rm
e
d
o
n
t
h
e VAR
r
e
s
i
d
u
al
s
fr
o
m
t
h
e VAR li
n
ea
r
filte
r
[
55
]
,
wh
ic
h
is bas
e
d
o
n
t
h
e
co
rr
e
latio
n
d
i
m
e
n
sio
n
o
f
t
h
e
n
o
n
li
n
e
a
r
st
ru
ct
ur
e
i
n
t
h
e
t
e
st ti
m
e
s
e
r
i
e
s. I
f
t
h
e te
s
t
n
e
g
ate
s
t
h
e
nu
ll
h
y
p
ot
h
e
s
i
s
, t
h
e te
s
te
d
ti
m
e
s
e
r
ie
s
i
s
n
o
n
li
n
ea
r
.
Ent
r
o
p
y
2025, 27, 345
5 o
f
26
t
Fi
na
lly,
w
e te
s
t t
h
e
n
o
n
li
n
e
a
r
G
r
an
ge
r
r
el
a
tio
n
b
et
w
ee
n
an
y t
w
o
n
o
d
e
s
as
t
h
e e
d
ge
o
f
t
h
e
n
et
w
o
r
k. T
h
e
n
u
ll
h
y
p
ot
h
e
s
i
s
(
H
0
)
i
s
t
h
at X
t
i
s
n
ot G
r
a
n
g
e
r
c
a
u
s
i
n
g
Y . I
n
p
r
a
c
ti
c
e,
co
n
d
itio
n
al i
n
d
e
p
e
n
d
e
n
ce i
s
te
s
te
d
us
i
n
g
fi
n
ite la
g
s
l
x
a
n
d
l
y
l
tt
tt
Y
+1
|(X
t
x
;
Y
l
y
) ∼ Y
+1
|Y
l
y
,
(
2
)
l
t
t
t
t
t
X,Y,
X,YY
(
x
,
y
)(
y
)
w
h
e
r
e X
t
s
= (X
t−l
x
+1
,
·
·
·
X
t
)
an
d
Y
l
s
= (Y
−l
y
+1
,
·
·
·
Y ), l
x
, l
y
> 1.
W
e
s
upp
o
s
e t
ha
t Z
t
=
Y
+1
a
n
d
z
t
∈ Z
t
. T
h
e
r
e
f
o
r
e, U = (X,Y, Z)
d
e
n
ote
s
a t
h
r
ee
-
va
r
iate
r
a
n
d
o
m
va
r
ia
b
le,
d
ist
r
ib
u
t
e
d
as U
t
= (X
t
,Y , Z
t
). T
h
e
n
, Eq
u
atio
n
(
2
)
ca
n
b
e
r
e
wr
itt
e
n
as
f
f
Y,Z
(
x
,
y
,
z
)
=
f
f
Z
(
y
,
z
)
.
Follo
w
i
n
g
[
54
]
, t
h
e
nu
ll
h
y
p
ot
h
e
s
i
s
ca
n
be i
mp
lie
d
a
s
q
≡ E
[f
X,Y,Z
(
x
,
y
,
z
)
f
Y
(
y
) −
f
X,Y
(
x
,
y
)
f
Y,Z
(
y
,
z
)
]
> 0.
(
3
)
T
h
e
n
, t
h
e te
s
t
s
tati
s
tic
f
o
r
n
o
n
li
n
ea
r
G
r
a
n
g
e
r
ca
us
ality i
s
n − 1
n
˜˜˜
T
n
(
ε
n
) =
n(n − 2)
∑
i=1
[f
X,Y,Z
(
x
i
,
y
i
,
z
i
)
f
Y
(
y
i
) −
f
X,Y
(
x
i
,
y
i
)
f
Y,Z
(
y
i
,
z
i
)
]
,
(
4
)
˜
w
h
e
r
e n i
s
t
h
e
s
ize o
f
sam
p
le,
sh
o
w
i
n
g t
h
e
n
u
mb
e
r
o
f
n
o
d
e
s
i
n
t
h
e
n
et
w
o
r
k,
an
d
f
U
(u
i
) i
s
t
h
e lo
c
al
d
e
ns
ity e
s
ti
m
ato
r
o
f
U at u
i
[
56
]
. T
h
e te
s
t
s
tati
s
ti
c
T
n
c
a
n
b
e i
n
te
rpr
ete
d
a
s
a
n
av
e
r
a
g
e
ov
e
r
local
B
D
S
t
e
st statistics
[
55
]
,
f
o
r
t
h
e
co
nd
itio
n
al
d
ist
r
ib
u
tio
n
o
f
X a
nd
Z,
g
iv
e
n
Y
=
y
i
. I
n
t
h
i
s
ca
s
e, t
h
e te
s
t
s
tati
s
tic T
n
i
s
a
s
y
mp
totically
n
o
rm
ally
d
i
s
t
r
ib
u
te
d
.
W
e obtai
n
t
h
e a
s
y
mm
et
r
ic
m
at
r
i
x
E = (e
i
j
)
n
×
n
t
hr
o
u
g
h
a
n
o
n
li
n
ea
r
G
r
a
n
g
e
r
ca
us
ality
te
s
t to
r
eveal t
h
e
d
i
r
ectio
n
o
f
s
e
n
ti
m
e
n
t
sp
illove
r
bet
w
ee
n
eve
r
y
p
ai
r
o
f
fi
rms
(
i,
j
)
, t
h
at i
s
,
(
1,
e
i
j
=
0,
i
f
i n
o
line
a
r
Gr
a
nge
r
c
a
uses
j
w
h
en i =
j
;
o
t
h
e
r
wise.
(
5
)
W
e
f
ollo
w
[
22
]
an
d
s
upp
o
s
e t
ha
t
f
o
r
e
ach
i ∈ (1,2
·
·
·
n), e
ii
= 0,
w
h
i
ch
ex
c
l
ud
e
s
t
h
e
co
nn
ectio
ns
bet
w
ee
n
t
h
e
n
o
d
e a
n
d
it
s
el
f
.
T
h
i
s
p
a
p
e
r
u
s
e
s
t
h
e
n
et
w
o
r
k
d
e
ns
ity
d
(
G
) to
m
e
as
ur
e t
h
e
c
o
nn
e
c
tio
ns
o
f
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t i
n
t
h
e
ma
r
ket. T
h
e
d
e
ns
ity
d
(
G
)
cha
r
ac
te
r
ize
s
h
o
w
c
lo
s
ely
s
to
c
k
n
o
d
e
s
a
r
e i
n
t
h
e
n
et
w
o
r
k. T
h
e l
a
r
ge
r
t
h
e
d
e
ns
ity o
f
t
h
e
n
et
w
o
r
k, t
h
e
m
o
r
e
c
lo
s
e
n
e
ss
b
et
w
ee
n
s
to
c
k
n
o
d
e
s
[
57
]
. T
h
e
d
e
ns
ity o
f
t
h
e
n
et
w
o
r
k ca
n
be calc
u
late
d
a
s
f
ollo
w
s
:
2L
d
(
G
) =
n(n − 1)
,
(
6
)
w
h
e
r
e L
r
e
pr
e
s
e
n
t
s
t
h
e a
c
t
u
al
n
u
mb
e
r
o
f
e
d
ge
s
c
o
nn
e
c
te
d
to t
h
e
n
et
w
o
r
k, a
n
d
n
d
e
n
ote
s
t
h
e
n
u
mb
e
r
o
f
n
o
d
e
s
. T
h
e val
u
e
r
a
n
g
e o
f
n
et
w
o
r
k
d
e
ns
ity i
s
[
0,1
];
t
h
e
n
et
w
o
r
k
d
e
ns
ity i
s
o
n
e
w
h
e
n
t
h
e
n
et
w
o
r
k i
s
fu
lly co
nn
ecte
d
.
2.1.2.
M
ea
sur
i
n
g
t
h
e Fi
rm’s
S
e
n
ti
m
e
n
t Co
nn
ecte
d
n
e
ss
:
A
n
E
n
t
r
o
p
y
W
ei
g
h
t
M
et
h
o
d
T
h
i
s
p
a
p
e
r
c
o
ns
i
d
e
r
s
t
h
e
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
o
f
fi
r
ms
bas
e
d
o
n
t
h
e
s
e
n
ti
m
e
n
t
s
p
illove
r
n
et
w
o
r
k. A
s
e
n
ti
m
e
n
t
s
p
illove
r
n
et
w
o
r
k
r
e
f
e
r
s
to t
h
e
c
o
nc
e
p
t t
ha
t
s
e
n
ti
m
e
n
t
can
s
pr
e
a
d
f
r
o
m
o
n
e i
n
d
ivi
du
a
l to
an
ot
h
e
r
w
it
h
i
n
a
s
o
c
i
a
l
n
et
w
o
r
k. It
s
u
gge
s
t
s
t
ha
t t
h
e
s
e
n
ti
m
e
n
t ex
pr
e
ss
e
d
b
y o
n
e i
n
d
ivi
du
a
l
can
i
n
fl
u
e
nc
e t
h
ei
r
n
eig
hb
o
r
’s
s
e
n
ti
m
e
n
t
s
t
a
te.
A
cc
o
rd
i
n
g to
Ma
tt
h
e
w
et
a
l.
(
2008
)
[
29
]
, i
n
a
fi
nanc
i
a
l
n
et
w
o
r
k
w
h
e
r
e
n
o
d
e
s
r
e
pr
e
s
e
n
t
fi
r
ms
an
d
li
n
k
a
ge
s
a
r
e fi
nanc
i
a
l
c
o
nn
e
c
tio
ns
s
u
ch
as
d
i
r
e
c
to
r
r
el
a
tio
nsh
i
p
s
[
24
]
an
d
o
b
li
-
g
a
tio
n
c
o
nn
e
c
tio
ns
[
23
]
, fi
r
ms
sh
o
u
l
d
ha
ve
h
ig
h
c
o
nn
e
c
te
d
n
e
ss
i
f
t
h
ey
a
r
e
(
1
)
li
n
ke
d
to
man
y ot
h
e
r
fi
r
ms
,
m
e
an
i
n
g t
h
ey
ha
ve
man
y
d
i
r
e
c
te
d
c
o
nn
e
c
tio
ns
w
it
h
ot
h
e
r
fi
r
ms
;
(
2
)
clo
s
e to ot
h
e
r
i
nd
ivi
du
al
s
,
m
ea
n
i
n
g
t
h
ey
h
ave le
ss
d
i
s
ta
n
ce a
nd
ca
n
spr
ea
d
i
nf
o
rm
atio
n
to ot
h
e
r
fi
r
ms
;
(
3
)
o
n
t
h
e
p
a
t
h
c
o
nn
e
c
ti
n
g
an
y ot
h
e
r
p
a
i
r
s
o
f
fi
r
ms
,
sh
o
w
i
n
g t
ha
t t
h
e
s
e
Ent
r
o
p
y
2025, 27, 345
6 o
f
26
fi
r
ms
can
t
r
ansm
it i
nf
o
r
ma
tio
n
e
as
ily;
an
d
(
4
)
li
n
ke
d
to ot
h
e
r
fi
r
ms
t
ha
t
ha
ve
man
y
c
o
n-
n
e
c
tio
ns
, i
n
d
i
ca
ti
n
g t
ha
t t
h
e
s
e fi
r
ms
a
r
e li
n
ke
d
to ot
h
e
r
a
u
t
h
o
r
ity
n
o
d
e
s
an
d
can
s
pr
e
a
d
i
nf
o
r
ma
tio
n
e
ff
e
c
tively. T
h
e
r
e
f
o
r
e, t
h
i
s
p
a
p
e
r
c
o
ns
i
d
e
r
s
f
o
ur
pr
i
ma
r
y ty
p
e
s
o
f
s
e
n
ti
m
e
n
t
n
e
t
w
o
r
k c
e
n
t
r
aliti
e
s
, i.
e
.,
d
e
g
r
ee
c
e
n
t
r
ality, clo
s
e
n
e
ss
c
e
n
t
r
ality, b
e
t
w
ee
nn
e
ss
c
e
n
t
r
ality, a
nd
ei
g
e
n
vecto
r
ce
n
t
r
ality.
T
h
e
d
e
g
r
ee ce
n
t
r
ality
(
DC
)
i
s
a
r
elatively i
n
t
u
itive i
n
d
e
x
, t
h
at i
s
, t
h
e
n
o
d
e
s
w
it
h
m
o
r
e
co
nn
ecte
d
e
dg
e
s
ca
n
s
p
r
ea
d
to
m
o
r
e fi
rms
.
S
upp
o
s
e t
h
at k
i
i
s
t
h
e
n
u
m
be
r
o
f
n
ei
g
h
bo
rs
o
f
n
o
d
e
i a
nd
n is t
h
e
total
num
b
e
r
o
f
n
o
d
e
s i
n
t
h
e
n
e
t
w
o
r
k
, t
h
e
n
t
h
e
d
e
g
r
ee
c
e
n
t
r
ality o
f
n
o
d
e
i
ca
n
be
m
ea
sur
e
d
a
s
DC
i
= k
i
/(n − 1).
(
7
)
T
h
e
c
lo
s
e
n
e
ss
c
e
n
t
r
a
lity
d
e
n
ote
s
t
h
e
a
ve
r
a
ge
sh
o
r
te
s
t
s
p
illove
r
d
i
s
t
anc
e
b
et
w
ee
n
a
n
o
d
e a
n
d
all ot
h
e
r
n
o
d
e
s
. T
h
e clo
s
e
n
e
ss
ce
n
t
r
ality
(
CC
i
)
o
f
n
o
d
e i i
s
calc
u
late
d
a
s
,
CC
i
= (n − 1)
∑
d
i
j
,
(
8
)
i=
j
i
n
∑
n
n
w
h
ile n i
s
t
h
e
num
be
r
o
f
i
n
t
h
e
n
et
w
o
r
k a
n
d
d
i
j
i
s
t
h
e
sh
o
r
te
s
t
d
i
s
ta
n
ce
fr
o
m
i to
j
.
T
h
e
b
et
w
ee
nn
e
ss
ce
n
t
r
ality
(B
C
)
m
ea
s
u
r
e
s
h
o
w
m
a
n
y
s
p
illove
r
p
at
hs
a
n
o
d
e lie
s
o
n
,
i
n
d
i
c
ati
n
g t
h
e i
m
p
o
r
ta
nc
e o
f
a
n
o
d
e i
n
s
pr
ea
d
i
n
g t
h
r
o
u
g
h
d
i
ff
e
r
e
n
t
p
a
r
t
s
o
f
t
h
e
n
et
w
o
r
k,
t
h
at i
s
,
B
C =
i
j
,
(
9
)
j
=1
i
j
wh
e
r
e
n
i
j
d
e
n
ot
e
s t
h
e
num
b
e
r
o
f
s
h
o
r
t
e
st
e
d
g
e
s
fr
o
m
i to
j
a
nd
n
i
j
is t
h
e
num
b
e
r
o
f
e
d
g
e
s
fr
o
m
i to
j
.
Ei
g
e
n
v
e
cto
r
c
e
n
t
r
ality
(
EC
i
)
i
nd
icat
e
s
t
h
e
i
mp
o
r
ta
n
c
e
o
f
a
n
o
d
e
i
n
a
sp
illov
e
r
n
e
t
w
o
r
k
;
s
p
e
cifically, t
h
e
co
nn
e
ct
e
dn
e
ss o
f
a
n
o
d
e
is
r
e
lat
e
d
to t
h
e
num
b
e
r
a
nd
p
o
w
e
r
o
f
t
h
e
a
d
jac
e
n
t
n
o
d
e
s
[
58
]
. I
f
x
i
i
s
t
h
e i
m
p
o
r
t
an
t
m
e
as
ur
e o
f
n
o
d
e i, t
h
e
c
e
n
t
r
a
lity i
s
ca
l
c
u
l
a
te
d
i
n
t
h
e
f
ollo
w
i
n
g
te
x
t
:
n
EC
i
= c
∑
a
i
j
x
j
,
(
10
)
j
=1
w
h
ile c i
s
a
pr
o
p
o
r
tio
n
al co
ns
ta
n
t.
a
i
j
= 1 i
f
ve
r
tice
s
i a
n
d
j a
r
e co
nn
ecte
d
by a
n
e
dg
e, a
n
d
a
i
j
= 0 i
f
t
h
ey a
r
e
n
ot.
To
m
itig
a
te t
h
e
b
i
as
an
d
viol
a
tio
n
i
n
m
e
as
ur
i
n
g
a
fi
r
m’s
c
o
nn
e
c
tio
ns
,
w
e
f
ur
t
h
e
r
q
u
a
n
ti
f
y a fi
rm’
s s
e
n
ti
m
e
n
t co
nn
e
ct
e
dn
e
ss by a
gg
r
e
g
ati
n
g
t
h
e
s
e
c
e
n
t
r
ality i
nd
e
x
e
s.
Supp
os
e
t
h
at C
i
j
r
e
pr
e
s
e
n
t
s
t
h
e ce
n
t
r
ality
m
ea
sur
e
m
e
n
t
j
f
o
r
co
mp
a
n
y i, t
h
at i
s
,
C
11
C
12
C
21
C
22
. .
C
n
1
C
n
2
C
13
C
14
C
23
C
24
(
11
)
..
C
n
3
C
n
4
T
h
e
n
, t
h
e e
n
t
r
o
p
y o
f
eac
h
f
acto
r
ca
n
i
s
calc
u
late
d
a
s
1
n
E
j
= −
l
n
n
∑
i=1
p
i
j
l
n
p
i
j
,
(
12
)
i
j
∑
n
C
i
j
w
h
e
r
e
p
=
C
i
j
, t
h
e
n
t
h
e
w
ei
g
h
t o
f
ce
n
t
r
ality i
n
d
e
x
j
ca
n
be calc
u
late
d
a
s
i=1
ω
=
4
1 − E
j
j
∑
k=1
1 − E
k
,
(
13
)
Ent
r
o
p
y
2025, 27, 345
7 o
f
26
T
h
e
w
e
i
g
h
t
ω
sh
o
ws
t
h
e
i
mp
o
r
ta
n
c
e
o
f
e
ac
h
c
e
n
t
r
ality i
n
a fi
rm
;
t
h
e
r
e
f
o
r
e
,
w
e
ca
n
m
e
a
sur
e
t
h
e fi
rm
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t co
nn
ecte
d
n
e
ss
a
s
4
Fi
r
mSenti
x
i
=
∑
j
=1
ω
j
C
i
j
,
(
14
)
w
h
e
r
e C
j
d
e
n
ote
s
t
h
e
n
et
w
o
r
k
c
e
n
t
r
a
lity
an
d
ω
j
i
s
t
h
e
w
eig
h
t o
b
t
a
i
n
e
d
f
r
o
m
t
h
e e
n
t
r
o
p
y
m
e
t
h
o
d
[
39
]
.
O
ur
s
e
n
ti
m
e
n
t co
nn
e
ct
e
dn
e
ss i
nd
e
x
ca
n
s
h
o
w
t
h
e
fu
ll s
p
illov
e
r
a
nd
co
nn
e
ctio
n
i
nf
o
r
ma
tio
n
o
f
a
fi
r
m
i
n
t
h
e
s
e
n
ti
m
e
n
t
n
et
w
o
r
k,
an
d
i
m
p
lie
s
t
ha
t
a
fi
r
m
w
it
h
a
h
ig
h
e
r
Fi
rmS
e
n
ti
x
h
a
s
m
o
r
e
s
e
n
ti
m
e
n
t
sp
illove
r
i
n
t
h
e
n
et
w
o
r
k.
2.2.
M
o
d
el I
d
enti
f
i
c
a
ti
o
n
a
n
d
V
a
r
i
ab
le
M
e
a
su
r
ements
T
h
i
s
p
a
p
e
r
c
h
oo
s
e
s
a fi
x
e
d-
e
ff
e
ct
m
o
d
e
l to i
d
e
n
ti
f
y t
h
e
i
mp
act o
f
s
e
n
ti
m
e
n
t co
nn
e
ct
e
d-
n
e
ss
o
n
c
r
a
sh
r
i
s
k. T
h
e
m
o
d
el ca
n
be e
x
pr
e
ss
e
d
a
s
CRASH
i,t
=γ
0
+ γ
1
CRASH
i,t−1
+ γ
2
Fi
r
mSenti
x
i,t
+ γ
3
DTURN
i,t
+ γ
4
R
O
E
i,t
i
+γ
5
MB
i,t
+ γ
6
LEV
,t
+ γ
7
SIZE
i,t
+ γ
8
O
paq
ue
i,t
+
θ
t
+ µ
i
+
ε
i,t
,
(
15
)
wh
e
r
e
CRASH
i,t
r
e
pr
e
s
e
n
t
s
t
h
e
c
r
a
sh
r
i
s
k
f
o
r
fi
rm
i i
n
y
e
a
r
t. Fi
r
mSenti
x
i,t
i
s
t
h
e
s
e
n
ti
m
e
n
t
co
nn
ecte
d
n
e
ss
o
f
fi
rm
i at yea
r
t.
θ
t
a
n
d
µ
i
a
r
e yea
r
a
n
d
fi
rm
d
umm
ie
s
.
Follo
w
i
n
g C
a
lle
n
an
d
F
an
g
(
2013
)
[
59
]
, E
q
u
a
tio
n
(
15
)
a
l
s
o i
nc
l
ud
e
s
v
a
r
io
u
s
m
u
lti
d
i
-
m
e
ns
io
n
al e
c
o
n
o
m
i
c
va
r
ia
b
le
s
a
s
c
o
n
t
r
ol va
r
ia
b
le
s
,
w
h
i
ch
a
r
e t
h
e lo
g
a
r
it
hm
i
c
total a
ss
et
s
(
SIZE
)
,
b
ook
-
to
-ma
r
ket
r
a
tio
(BM)
,
ass
et–li
ab
ility
r
a
tio
(
LEV
)
,
an
d
r
et
ur
n
o
n
n
et
ass
et
s
(
R
O
E
)
, as
w
e
ll as co
mp
a
n
y
-
s
p
e
cific
f
acto
r
s s
u
c
h
as i
nf
o
rm
atio
n
o
p
acity i
nd
icato
r
s
(
O
paq
ue
)
a
n
d
i
n
ve
s
to
r
d
ive
r
g
e
n
ce
s
t
r
e
n
g
t
h
(
DTRUN
)
. A
dd
itio
n
ally, to
m
iti
g
ate t
h
e
p
ote
n
tial la
g
e
f-
f
e
ct ca
u
s
e
d
by c
r
as
h
r
is
k
,
w
e
i
n
co
rp
o
r
at
e
t
h
e
la
g
ph
as
e
o
f
c
r
as
h
r
is
k
as a
n
a
dd
itio
n
al co
n
t
r
ol
va
r
iabl
e
. As t
h
e
t
r
a
d
i
n
g
a
nd
s
e
n
ti
m
e
n
t
m
ay b
e
i
n
fl
u
e
n
c
e
d
by
h
i
g
h-fr
e
q
u
e
n
cy t
r
a
d
i
n
g
(
H
F
T
)
,
w
e al
s
o co
ns
i
d
e
r
t
h
e HFT
m
a
r
ket
s
ize
(
HFT
)
i
n
t
h
i
s
p
a
p
e
r
. T
h
e
d
efi
n
itio
ns
o
f
t
h
e va
r
iable
s
us
e
d
i
n
t
h
i
s
s
t
u
d
y a
r
e
pr
ovi
d
e
d
i
n
A
pp
e
n
d
i
x
A.
Acco
rd
i
n
g
to Ki
m
e
t al.
(
2014
)
[
60
]
,
w
e
co
n
si
d
e
r
t
w
o c
r
as
h
r
is
k
m
e
as
ur
e
s.
W
e
calc
u
lat
e
fi
r
m-s
p
e
c
ifi
c
w
eekly
r
et
ur
ns
(
W
j
,τ
)
b
y t
a
ki
n
g t
h
e
na
t
ur
a
l log
a
r
it
hm
o
f
o
n
e
a
dd
e
d
to t
h
e
r
e
s
i
d
u
al ea
rn
i
n
g
s
fr
o
m
t
h
e e
x
te
n
d
e
d
m
a
r
ket
m
o
d
el
r
e
g
r
e
ss
io
n
f
o
r
eac
h
yea
r
a
n
d
fi
rm
.
r
j
,τ
=
α
j
+ β
1
j
r
m,τ−2
+ β
2
j
r
m,τ−1
+ β
3
j
r
m,τ
+ β
4
j
r
m,τ+1
+ β
5
j
r
m,τ+2
+
ε
j
,τ
,
(
16
)
wh
e
r
e
r
j
,τ
is t
h
e
r
e
t
urn
o
n
stoc
k
j
i
n
w
ee
k
τ.
r
m,τ−2
,
r
m,τ−1
,
r
m,τ
,
r
m,τ+1
, a
nd
r
m,τ+2
r
e
pr
e
s
e
n
t t
h
e
m
a
r
ket
r
et
u
rn
i
n
w
eek τ − 2, τ − 1, τ, τ + 1, a
n
d
τ + 2,
r
e
s
p
e
c
tively. T
h
e lea
d
a
n
d
la
g
te
rms
f
o
r
t
h
e
m
a
r
ket i
n
d
e
x
r
et
urn
a
r
e i
n
cl
u
d
e
d
to allo
w
f
o
r
n
o
n-s
y
n
c
hr
o
n
o
us
t
r
a
d
i
n
g
[
60
]
.
T
h
e fi
r
m-s
p
e
c
ifi
c
w
eekly
r
et
ur
ns
f
o
r
fi
r
m
j
i
n
w
eek τ
a
r
e
c
o
m
pu
te
d
b
y W
j
,τ
= 1 +
l
n
(1 +
ε
j
,τ
),
w
h
e
r
e
ε
i
s
t
h
e
r
e
s
i
du
al i
n
E
q
u
atio
n
(
16
)
. T
h
e
n
,
w
e
s
ele
c
t
n
e
g
ative
c
o
n
d
itio
n
al
r
et
u
rn
s
ke
w
n
e
ss
(
NCSKEW
)
a
s
a
m
ea
s
u
r
e o
f
ea
ch
c
o
m
p
a
n
y
’s
s
to
c
k
p
r
i
c
e
r
i
s
k, fi
gu
r
e
d
b
y
taki
n
g
t
h
e
n
e
g
ative o
f
t
h
e t
h
i
r
d
m
o
m
e
n
t o
f
fi
rm-sp
ecific
w
eekly
r
et
urns
f
o
r
eac
h
yea
r
a
n
d
d
ivi
d
i
n
g it
b
y t
h
e
c
u
b
i
c
s
ta
n
d
a
rd
d
eviatio
n
o
f
fi
r
m-s
p
e
c
ifi
c
w
eekly
r
et
ur
ns
. I
n
p
a
r
ti
c
u
la
r
,
w
e co
mpu
te t
h
e (NCSKEW)
f
o
r
eac
h
fi
rm
j
i
n
yea
r
t a
s
j
,t
T
3
T
2
3/2
NCSKEW=
−T(T − 1)
3/2
∑
τ=1
W
j
,τ
,
(
17
)
(T − 1)(T − 2)(
∑
τ=1
W
j
,τ
)
w
h
e
r
e T i
s
t
h
e
num
be
r
o
f
w
eekly
r
et
urns
d
ur
i
n
g
yea
r
t. A
h
i
g
h
e
r
NCSKEW i
n
d
icate
s
t
h
at
a
s
to
c
k i
s
m
o
r
e
pr
o
n
e to
c
r
ash
e
s
,
an
d
vi
c
e ve
r
sa
. A
n
ot
h
e
r
m
e
as
ur
e o
f
c
r
ash
r
i
s
k i
s
t
h
e
d
o
w
n-
to
-
up
vol
a
tility
(
DUV
O
L
)
. To
ca
l
c
u
l
a
te t
h
i
s
m
e
as
ur
e,
w
e
f
ollo
w
[
12
]
an
d
s
e
p
a
r
a
te
t
h
e
w
eek
s
w
it
h
fi
r
m-s
p
e
c
ifi
c
w
eekly
r
et
ur
ns
ab
ove t
h
e
ann
u
a
l
a
ve
r
a
ge
(
up
w
eek
s)
f
r
o
m
Ent
r
o
p
y
2025, 27, 345
8 o
f
26
t
h
o
s
e
w
it
h
fi
rm-sp
ecific
r
et
u
rns
belo
w
t
h
e a
nn
u
al
m
ea
n
(
d
o
w
n
w
eek
s)
.
W
e t
h
e
n
co
mp
u
te
t
h
e
s
ta
n
d
a
r
d
d
eviatio
n
f
o
r
eac
h
su
b
s
et o
f
w
eek
s
. It i
s
calc
u
late
d
a
s
u
n − 1)
∑
W
2
j
,
∑
u
∈
u
p
W
2
DUV
O
L
j
,t
= l
n
(
(n
d
− 1)
d∈d
o
w
n
j
,
u
d
,
(
18
)
wh
e
r
e
W
j
,τ
r
e
pr
e
s
e
n
ts t
h
e
w
ee
k
ly
r
e
t
urn
s to stoc
k
i
dur
i
n
g
p
e
r
io
d
τ. n
u
is t
h
e
num
b
e
r
o
f
up
w
eek
s
a
n
d
n
d
i
s
t
h
e
num
be
r
o
f
d
o
w
n
w
eek
s
.
3. Data
O
ur
r
e
s
e
a
r
ch
sam
p
le i
nc
l
ud
e
s
t
h
e
d
a
ily t
ur
n
ove
r
o
f
s
to
c
k
s
c
o
m
pr
i
s
i
n
g t
h
e
S&P
500 i
nd
e
x
fr
o
m
4 Ja
nu
a
r
y 2006 to 31 D
e
c
e
m
b
e
r
2021.
F
ollo
w
i
n
g
H
u
y
nh
e
t al.
(
2021
)
[
61
]
,
w
e
a
pp
ly
s
p
e
c
ifi
c
cr
ite
r
ia to
s
ele
c
t
s
to
c
k
s
f
o
r
o
u
r
s
t
ud
y
:
(
i
)
ex
c
l
u
s
io
n
o
f
s
to
c
k
s
s
u
s
p
e
n
d
e
d
f
o
r
m
o
r
e t
h
a
n
180 co
ns
ec
u
tive t
r
a
d
i
n
g
d
ay
s
;
(
ii
)
r
e
m
oval o
f
un
li
s
te
d
s
tock
s
at t
h
e be
g
i
nn
i
n
g
o
f
t
h
e
s
t
ud
y
p
e
r
io
d
;
an
d
(
iii
)
eli
m
i
na
tio
n
o
f
d
eli
s
te
d
s
to
c
k
s
a
t t
h
e e
n
d
o
f
t
h
e
sam
p
le
p
e
r
io
d
.
Co
ns
e
qu
e
n
tly, o
ur
a
n
aly
s
i
s
f
oc
us
e
s
o
n
a total o
f
372 co
ns
tit
u
e
n
t
s
tock
s
o
f
t
h
e
S&
P
500. All
t
h
e
d
ata
us
e
d
i
n
t
h
i
s
s
t
u
d
y a
r
e
fr
o
m
t
h
e
W
IND
d
ataba
s
e.
S
i
m
il
a
r
to
Ba
ke
r
et
a
l.
(
2012
)
[
20
]
,
w
e
ch
oo
s
e t
h
e t
ur
n
ove
r
r
a
te o
f
a
fi
r
m
as
o
ur
pr
i
ma
r
y
s
e
n
ti
m
e
n
t
pr
oxy.
Ba
ke
r
an
d
S
tei
n
(
2004
)
[
51
]
ass
e
r
t t
ha
t
sh
o
r
t
-sa
le
c
o
ns
t
r
a
i
n
t
s
ca
u
s
e
n
ois
e
t
r
a
d
e
r
s to
pr
e
f
e
r
e
n
tially t
r
a
n
sact
wh
e
n
f
ee
li
n
g
o
p
ti
m
istic
r
at
h
e
r
t
h
a
n
p
e
ssi
m
istic.
Sp
e
cifically, i
rr
atio
n
al i
n
v
e
sto
r
s
’
o
p
ti
m
is
m
is
r
e
fl
e
ct
e
d
i
n
a
n
asc
e
nd
i
n
g
t
urn
ov
e
r
r
atio,
wh
il
e
t
h
ei
r
p
e
ss
i
m
i
sm
co
rr
e
sp
o
n
d
s
to a
d
e
s
ce
n
d
i
n
g
t
urn
ove
r
r
atio. T
h
i
s
t
h
eo
r
etical li
n
k bet
w
ee
n
s
e
n
ti
m
e
n
t a
nd
t
r
a
d
i
n
g
activity
h
as also b
ee
n
s
upp
o
r
t
e
d
e
mp
i
r
ically
[
50
]
. T
h
e
t
urn
ov
e
r
r
atio i
s
a
l
s
o
w
ell
s
u
ite
d
f
o
r
ca
p
t
ur
i
n
g vol
a
tility
an
d
t
h
e
m
e
an
–v
a
r
i
anc
e
r
el
a
tio
nsh
i
p
du
e to it
s
d
a
ily
a
v
a
il
ab
ility. A
s
a
w
i
d
ely
acc
e
ss
i
b
le
m
et
r
i
c
c
o
m
pu
te
d
d
i
r
e
c
tly
f
r
o
m
s
to
c
k ex
chan
ge
d
ata, t
h
e t
urn
ove
r
r
atio e
x
p
licitly
m
ea
sur
e
s
s
e
n
ti
m
e
n
t i
n
e
qu
ity
m
a
r
ket
s
r
at
h
e
r
t
h
a
n
ot
h
e
r
fi
n
a
n
cial
m
a
r
ket
s
[
62
]
.
Table 1
r
e
p
o
r
t
s
t
h
e
d
e
s
c
r
i
p
tive
s
tati
s
tic
s
f
o
r
key va
r
iable
s
i
n
t
h
i
s
s
t
ud
y.
S
a
m
p
le fi
rms
ex
h
i
b
it
a
m
e
an
t
ur
n
ove
r
o
f
0.9509,
m
e
d
i
an
o
f
0.6812,
an
d
ma
xi
m
u
m
v
a
l
u
e o
f
110.7406,
s
u
gge
s
ti
n
g
s
ig
n
ifi
can
t vol
a
tility i
n
ma
r
ket
s
e
n
ti
m
e
n
t
w
it
h
i
n
t
h
e U.
S
. e
q
u
ity
ma
r
ket. T
h
e
a
ve
r
a
ge
c
r
ash
r
i
s
k
m
e
as
ur
e
(
NCSKEW
i,t
)
s
t
an
d
s
a
t 0.5845,
a
lig
n
i
n
g
w
it
h
pr
io
r
e
m
p
i
r
i
ca
l
fi
nd
i
n
g
s. T
h
e
s
e
r
e
s
u
lts
h
i
g
h
li
g
h
t t
h
e
e
l
e
vat
e
d
c
r
as
h
r
is
k
a
m
o
n
g
S&P
500 fi
rm
s,
und
e
r
sco
r
i
n
g
t
h
ei
r
r
eleva
n
ce
f
o
r
i
n-
d
e
p
t
h
i
n
ve
s
ti
g
atio
n
.
Ta
b
le 1. De
s
c
r
i
p
tive
s
tati
s
tic
s
o
f
all va
r
iable
s
.
Va
r
ia
b
le
M
ea
n
M
e
d
ia
n
St
d
.Dev
25%
75%
T
urn
ove
r
Fi
rmS
e
n
ti
x
NC
S
KE
W
DUV
O
L
S
YNC
O
p
a
qu
e
C
s
co
r
e
DTRUN
R
O
E
LEV
S
IZE
MB
HFT
0.9509
0.3488
0.5845
0.3730
0.9430
0.0852
0.1024
−0.0175
25.9194
0.6290
23.4522
0.3072
21.7037
0.6812
0.3448
0.3264
0.2500
0.7374
0.0690
0.0598
−0.2066
14.6409
0.6195
23.4596
0.0053
7.5198
1.2750
0.0518
1.4818
1.0784
1.4826
0.0640
0.5051
9.8562
887.9436
0.2739
1.9902
1.3380
20.9112
0.4537
0.3139
−0.2292
−0.2560
0.1605
0.0396
−0.0251
−2.9545
8.7548
0.4758
22.4874
−0.3439
6.6078
1.0668
0.3815
1.0159
0.8145
1.3944
0.1125
0.1407
2.6064
24.1418
0.7622
24.4137
0.7442
39.5887
Not
e:
T
h
is tabl
e
r
e
p
o
r
ts t
h
e
m
e
a
n
,
m
e
d
ia
n
, sta
nd
a
rd
d
e
viatio
n
, a
nd
upp
e
r
a
nd
lo
w
e
r
q
u
a
r
til
e
s o
f
t
h
e
d
ata
u
s
e
d
f
o
r
p
a
n
el
r
e
g
r
e
ss
io
n
a
n
d
n
et
w
o
r
k co
ns
t
ru
ctio
n
. All va
r
iable
s
i
n
t
h
i
s
table a
r
e
d
efi
n
e
d
i
n
A
pp
e
n
d
i
x
A.
Ent
r
o
p
y
2025, 27, 345
9 o
f
26
4. E
m
p
i
r
ical Re
s
u
lt
s
4.1. Sentiment S
p
ill
o
v
e
r
Netw
o
r
k Am
o
ng S
&
P
500 St
o
cks
W
e
co
n
st
ru
ct a s
e
n
ti
m
e
n
t s
p
illov
e
r
n
e
t
w
o
r
k
b
e
t
w
ee
n
S&P
500 stoc
k
s
fr
o
m
2006 to 2021.
F
i
g
ur
e
1 s
h
o
w
s t
h
e
n
e
t
w
o
r
k
i
n
t
h
e
y
e
a
r
s 2007, 2008, 2019, a
nd
2020. I
n
e
ac
h
n
e
t
w
o
r
k
,
n
o
d
e
s
r
e
pr
e
s
e
n
t fi
rms
a
n
d
d
i
r
ecte
d
a
rr
o
w
s
sh
o
w
t
h
e
sp
illove
r
bet
w
ee
n
fi
rms
.
Fig
ur
e 2
a
d
e
p
i
c
t
s
t
h
e
c
u
m
u
l
a
tive
d
i
s
t
r
i
b
u
tio
n
o
f
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
f
o
r
t
h
e
ye
a
r
s
2008, 2009, 2019,
an
d
2020. T
h
e
r
e
s
u
lt
s
r
eve
a
l
a
h
ig
h
ly
h
ete
r
oge
n
eo
u
s
d
i
s
t
r
i
b
u
tio
n
p
att
e
rn
,
wh
e
r
e
m
ost
n
e
t
w
o
r
k
n
o
d
e
s
e
x
h
ibit li
m
it
e
d
s
e
n
ti
m
e
n
t li
n
k
a
g
e
s
wh
il
e
a s
m
all s
u
bs
e
t
d
e
m
o
ns
t
r
at
e
s
s
i
g
n
ifica
n
t co
nn
e
ctivity. Notably, c
r
i
s
i
s
p
e
r
io
ds
(
2008 a
nd
2020
)
d
i
sp
lay
m
o
r
e
pr
o
n
o
u
n
ce
d
h
ete
r
o
g
e
n
eity a
n
d
co
nn
ecte
d
n
e
ss
co
mp
a
r
e
d
to
n
o
n-
c
r
i
s
i
s
yea
rs
. T
h
i
s
fi
n
d
i
n
g
i
s
vi
s
u
ally
s
upp
o
r
te
d
b
y t
h
e flatte
r
d
i
s
t
r
i
b
u
tio
n
c
u
r
ve
s
o
bs
e
r
ve
d
du
r
i
n
g
cr
i
s
i
s
p
e
r
io
d
s
, a
s
o
pp
o
s
e
d
to t
h
e
s
tee
p
e
r
d
i
s
t
r
ib
u
tio
ns
s
ee
n
i
n
2007 a
n
d
2019.
It i
s
i
mp
o
r
ta
n
t to
n
ote t
h
at o
ur
s
e
n
ti
m
e
n
t
n
et
w
o
r
k
s
a
r
e
d
r
ive
n
by e
x
o
g
e
n
o
us
i
n
ve
s
to
r
be
h
avio
r
al
f
acto
rs
r
at
h
e
r
t
h
a
n
co
mp
a
n
y
fun
d
a
m
e
n
tal
s
. T
h
i
s
d
i
s
ti
n
g
u
i
sh
e
s
o
ur
s
t
u
d
y
fr
o
m
pr
evio
u
s
r
e
s
e
a
r
ch
o
n
fi
nanc
i
a
l
n
et
w
o
r
k
s
t
ha
t
f
o
c
u
s
e
d
o
n
b
u
s
i
n
e
ss
c
o
nn
e
c
tio
ns
[
24,31
]
o
r
pr
ice li
n
ka
g
e
[
22,28
]
. F
r
o
m
Fi
g
ur
e 1,
w
e fi
n
d
t
h
at t
h
e co
nn
ecte
d
n
e
ss
o
f
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t
c
a
nn
ot
b
e
s
olely att
r
i
b
u
te
d
to
c
o
m
p
a
n
y fi
n
a
nc
ial
ch
a
r
a
c
te
r
i
s
ti
cs
. Fo
r
exa
m
p
le,
w
it
h
i
n
t
h
e
s
e
n
ti
m
e
n
t
n
e
t
w
o
r
k
, Am
a
z
o
n is i
n
fl
u
e
n
c
e
d
by
num
e
r
o
u
s ot
h
e
r
fi
rm
s,
wh
e
r
e
as J
P
M
o
r
g
a
n C
ha
se
a
n
d
G
o
ldm
a
n
S
a
chs,
d
e
s
p
ite
b
ei
n
g
g
lo
b
ally
r
e
c
o
g
n
ize
d
a
s
i
m
p
o
r
ta
n
t fi
n
a
nc
ial i
ns
tit
u
tio
ns
,
r
e
c
eive
f
e
w
e
r
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
f
r
o
m
ot
h
e
r
fi
r
ms
. T
h
i
s
p
h
e
n
o
m
e
n
o
n
h
ig
h
lig
h
t
s
t
h
e
p
ote
n
ti
a
l
f
o
r
a
fi
r
m
to e
s
t
ab
li
sh
c
o
nn
e
c
tio
ns
t
h
r
o
u
g
h
it
s
i
n
fl
u
e
nc
e o
n
an
ot
h
e
r
fi
r
m’s
s
e
n
ti
m
e
n
t, eve
n
w
it
h
o
u
t
d
i
r
e
c
tly
aff
e
c
ti
n
g o
w
n
e
r
sh
i
p
o
r
ass
et flo
w
s
. It
s
u
gge
s
t
s
t
ha
t
s
e
n
ti
m
e
n
t co
nn
ecte
d
n
e
ss
i
s
n
ot
s
olely
d
ete
rm
i
n
e
d
by t
h
e
pr
o
m
i
n
e
n
ce o
f
a fi
rm
.
(
a
)(
b
)(
c
)(
d
)
(
a
)
Fig
u
r
e 1. T
h
e
s
t
ru
c
t
ur
a
l
cha
r
ac
te
r
i
s
ti
c
ana
ly
s
i
s
o
f
s
e
n
ti
m
e
n
t
s
p
illove
r
n
et
w
o
r
k
s
.
(a)
sh
o
w
s
t
h
e
s
e
n
t
i
m
e
n
t
s
p
illov
e
r
n
et
w
o
r
k
i
n
2007
;
(b)
s
h
o
w
s
t
h
e
s
e
n
t
i
m
e
n
t
s
p
illov
e
r
n
et
w
o
r
k
i
n
2008
;
(
c
)
s
h
o
w
s
t
h
e
s
e
n
ti
m
e
n
t
sp
illove
r
n
et
w
o
r
k i
n
2019
;
(
d
)
sh
o
w
s
t
h
e
s
e
n
ti
m
e
n
t
sp
illove
r
n
et
w
o
r
k i
n
2020.
0.35 0.40 0.45
0.65 0.70
0.50 0.55 0.60
l
o
g(x*10)
1.5
1.0
0.5
0.0
0.5
1.0
l
o
g[
Pr
(X x)*10]
2007
2008
2019
2020
2006 2008
2010 2012 2014 2016 2018 2020
t
ime
0.04
0.05
0.06
0.07
0.08
0.09
0.10
0.11
ne
t
w
or
k densi
t
y
(
b
)
F
i
g
ur
e 2. T
h
e
s
t
ru
c
t
ur
al c
h
a
r
ac
te
r
is
t
ic o
f
s
e
n
t
i
m
e
n
t
s
p
illov
e
r
n
et
w
o
r
k
s.
(
a
)
is
t
h
e
c
umu
la
t
iv
e
d
is
t
r
ib
u
t
io
n
o
f
s
e
n
ti
m
e
n
t co
nn
e
ct
e
d
n
e
ss
f
o
r
t
h
e
y
e
a
rs
2008, 2009, 2019, a
n
d
2020
;
(
b
)
i
s
t
h
e
y
e
a
r
ly
n
e
t
w
o
r
k
d
e
ns
ity.
Ent
r
o
p
y
2025, 27, 345
10 o
f
26
Fig
ur
e 2
b
ex
am
i
n
e
s
t
h
e te
m
p
o
r
a
l evol
u
tio
n
o
f
n
et
w
o
r
k
d
e
ns
ity to
cha
r
ac
te
r
ize
s
y
s
te
m
i
c
i
n
te
r
c
o
nn
e
c
te
d
n
e
ss
. Not
ab
ly,
c
r
i
s
i
s
p
e
r
io
d
s
ex
h
i
b
it
pr
o
n
o
u
nc
e
d
s
p
ike
s
i
n
s
e
n
ti
m
e
n
t
n
et
w
o
r
k
d
e
n
sity. D
ur
i
n
g
t
h
e
2008 s
u
b
pr
i
m
e
c
r
isis,
d
e
n
sity
r
os
e
s
h
a
rp
ly
fr
o
m
0.0578 to 0.1159,
r
e
ac
h
i
n
g
t
h
e
h
i
g
h
e
st val
u
e
obs
e
r
v
e
d
ov
e
r
t
h
e
e
n
ti
r
e
sa
mp
l
e
p
e
r
io
d
.
F
ollo
w
i
n
g
t
h
is
p
e
a
k
,
d
e
n
sity s
u
bs
e
q
u
e
n
tly
d
e
c
li
n
e
d
s
tea
d
ily
u
n
til 2011,
w
h
e
n
t
h
e U.
S
.
s
ove
r
ei
g
n
d
e
b
t
c
r
i
s
i
s
t
r
i
gg
e
r
e
d
a
n
ot
h
e
r
s
i
g
n
i
f
i
c
a
n
t
i
n
c
r
ea
s
e. Co
n
ve
rs
ely,
n
et
w
o
r
k
d
e
ns
ity e
x
p
e
r
ie
n
ce
d
r
a
p
i
d
co
n
t
r
actio
n
d
ur
i
n
g
p
o
s
t
-
c
r
i
s
i
s
m
a
r
ket
r
ecove
r
y
ph
a
s
e
s
,
r
e
f
lecti
n
g
r
e
d
u
ce
d
i
nf
o
rm
atio
n
d
i
ffus
io
n
ac
r
o
ss
m
a
r
ket
p
a
r
tici
p
a
n
t
s
.
F
ur
t
h
e
rm
o
r
e,
d
ur
i
n
g
su
b
s
eq
u
e
n
t c
r
i
s
e
s
su
c
h
a
s
t
h
e
S
i
n
o
-
U
S
t
r
a
d
e
w
a
r
i
n
2018 a
n
d
t
h
e
COVID
-
19
c
r
i
s
i
s
i
n
2020, t
h
e
n
et
w
o
r
k
d
e
ns
ity
r
e
ach
e
d
v
a
r
yi
n
g
d
eg
r
ee
s
o
f
p
e
a
k
s
. He
nc
e,
w
e
can
i
nf
e
r
t
ha
t
n
et
w
o
r
k
d
e
ns
ity
p
o
ss
e
ss
e
s
a
c
e
r
t
a
i
n
level o
f
i
d
e
n
tifi
ca
tio
n
f
o
r
fi
nanc
i
a
l
c
r
is
e
s. T
h
is is b
e
ca
u
s
e
a
n
i
n
c
r
e
as
e
i
n
n
e
t
w
o
r
k
d
e
n
sity si
g
n
ifi
e
s
h
e
i
g
h
t
e
n
e
d
stoc
k
co
rr
e
latio
n
s,
t
h
e
r
eby i
n
c
r
ea
s
i
n
g
t
h
e likeli
h
oo
d
o
f
r
i
s
k t
r
a
nsm
i
ss
io
n
.
4.2.
Q
u
a
nti
f
y
ing t
h
e Im
pa
ct
o
f
Sentiment S
p
ill
o
v
e
r
o
n C
r
a
s
h
Risk
W
e
e
x
a
m
i
n
e
t
h
e
i
mp
act o
f
t
h
e
s
e
n
ti
m
e
n
t co
nn
e
ct
e
dn
e
ss
o
n
s
toc
k
pr
ic
e
c
r
a
sh
r
i
s
k
us
i
n
g
s
e
v
e
r
al
p
a
n
e
l
r
e
g
r
e
ssio
n
m
o
d
e
ls. To t
e
st
f
o
r
p
ossibl
e
colli
n
e
a
r
ity
pr
obl
e
m
s b
e
t
w
ee
n
e
x
p
la
n
a
-
to
r
y v
a
r
i
ab
le
s
,
a
c
o
rr
el
a
tio
n
ana
ly
s
i
s
i
s
ca
rr
ie
d
o
u
t. T
h
e
r
e
s
u
lt
s
sh
o
w
t
ha
t t
h
e
c
o
rr
el
a
tio
n
coe
f
ficie
n
t
s
bet
w
ee
n
va
r
iable
s
a
r
e le
ss
t
h
a
n
0.2, i
n
d
icati
n
g
t
h
at t
h
e
r
e
g
r
e
ss
io
n
m
o
d
el
d
oe
s
n
ot
h
ave colli
n
ea
r
ity
pr
oble
ms
. F
ur
t
h
e
rm
o
r
e,
s
i
n
ce t
h
e e
mp
i
r
ical lite
r
at
ur
e o
n
f
o
r
eca
s
ti
n
g
c
r
ash
r
i
s
k i
s
r
el
a
tively
n
e
w
, o
ur
ana
ly
s
i
s
ma
y o
m
it
s
o
m
e
c
r
ash
d
ete
r
m
i
nan
t
s
c
o
rr
el
a
te
d
w
it
h
ot
h
e
r
i
n
cl
ud
e
d
va
r
iabl
e
s
fr
o
m
t
h
e
r
e
g
r
e
ss
io
ns
. To
m
iti
g
at
e
p
ot
e
n
tial
pr
obl
e
ms
t
h
at ca
n
a
r
i
s
e
f
r
o
m
c
o
rr
el
a
te
d
ly o
m
itte
d
v
a
r
i
ab
le
s
,
w
e
a
tte
m
p
t to
u
s
e
a
fixe
d
e
ff
e
c
t
m
o
d
el.
B
e
f
o
r
e
it,
w
e
p
e
r
f
o
r
m
t
h
e H
a
u
sman
te
s
t,
w
h
i
ch
sh
o
w
s
t
ha
t t
h
e
m
o
d
el i
s
r
e
as
o
nab
le
[
63,64
]
. T
h
e
r
e
g
r
e
ss
io
n
r
e
su
lt
s
a
r
e
r
e
n
d
e
r
e
d
i
n
Table 2.
Ta
b
le 2. T
h
e
r
e
g
r
e
ss
io
n
o
f
s
e
n
ti
m
e
n
t co
nn
ecte
d
n
e
ss
a
n
d
c
r
a
sh
r
i
s
k.
Va
r
ia
b
le
DTURN
i,t
R
O
E
i,t
MB
i,t
i
LEV
,t
SIZE
i,t
O
paq
ue
i,t
0.0025
(
1.62
)
9.46 × 10
−6
***
(
2.72
)
−0.3461 ***
(-
32.36
)
0.1021
(
0.92
)
0.0233
(
1.25
)
0.3279
(
1.00
)
NCSKEW
i,t−1
0.0031
(
1.39
)
1.08 × 10
−5
(
1.33
)
−0.3556 ***
(
−26.93
)
0.1128
(
0.72
)
0.0350
(
1.18
)
0.6218
(
1.28
)
−0.1096 ***
(
−8.66
)
Col
u
m
n
(
1
)
NCSKE
W
i,
t
Col
u
m
n
(
2
)
NCSKE
W
i,
t
Col
u
m
n
(
3
)
DU
V
OL
i,
t
Col
u
m
n
(
4
)
DU
V
OL
i,
t
Fi
r
mSenti
x
i,t
0.5287 *
(
1.82
)
0.0025
(
1.63
)
9.98 × 10
−6
***
(
2.88
)
−0.3458 ***
(
−32.30
)
0.1013
(
0.93
)
0.0224
(
1.20
)
0.3393
(
1.00
)
1.2895 ***
(
2.82
)
0.0032
(
1.41
)
1.21 × 10
−5
(
1.53
)
−0.3547 ***
(
−26.83
)
0.1107
(
0.72
)
0.0328
(
1.11
)
0.6249
(
1.27
)
−0.1073 ***
(
−8.58
)
DUV
O
L
i,t−1
C
o
nst
a
nt
0.0173
(
0.02
)
−0.3993
(
−0.56
)
−0.1173 ***
(
−11.04
)
−0.1391
(
−0.30
)
−0.1167 ***
(
−10.99
)
−0.9980
(
−0.80
)
Fi
r
m
Ye
a
r
Ye
s
Ye
s
Ye
s
Ye
s
Ye
s
Ye
s
Ye
s
Ye
s
F
A
d
j
.R
2
O
b
s.
145.92
0.242
5564
141.38
0.243
5564
161.73
0.295
5564
153.33
0.298
5564
Note: T
h
i
s
t
ab
le
pr
e
s
e
n
t
s
t
h
e key
r
e
s
u
lt
s
f
r
o
m
t
h
e e
s
ti
ma
tio
n
o
f
t
h
e
r
el
a
tio
n
b
et
w
ee
n
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
an
d
s
to
c
k
pr
i
c
e
c
r
ash
r
i
s
k. Col
u
mns
(
1
)
an
d
(
3
)
a
r
e
r
e
s
u
lt
s
w
it
h
o
u
t t
h
e
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
v
a
r
i
ab
le,
an
d
col
umns
(
2
)
a
n
d
(
4
)
a
r
e
r
e
g
r
e
ss
io
n
r
e
su
lt
s
w
it
h
t
h
e
s
e
n
ti
m
e
n
t co
nn
ecte
d
n
e
ss
. T
h
e t te
s
t val
u
e
s
a
r
e i
n
p
a
r
e
n
t
h
e
s
e
s
,
w
h
e
r
e ***
an
d
* i
n
d
i
ca
te
s
t
a
ti
s
ti
ca
l
s
ig
n
ifi
canc
e
a
t t
h
e 1
%
an
d
10
%
level
s
,
r
e
s
p
e
c
tively. Det
a
il
s
r
eg
a
rd
i
n
g e
ach
va
r
iable
us
e
d
i
n
t
h
i
s
s
t
u
d
y a
r
e
pr
e
s
e
n
te
d
i
n
A
pp
e
n
d
i
x
A.
Ent
r
o
p
y
2025, 27, 345
11 o
f
26
A
s
ill
u
s
t
r
a
te
d
i
n
T
ab
le 2, t
h
e
r
eg
r
e
ss
io
n
c
oe
f
fi
c
ie
n
t
r
eve
a
l
s
a
p
o
s
itive
ass
o
c
i
a
tio
n
b
e
-
t
w
ee
n
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
an
d
s
to
c
k
pr
i
c
e
c
r
ash
r
i
s
k,
s
u
gge
s
ti
n
g t
ha
t
s
e
n
ti
m
e
n
t
co
nn
ecte
d
n
e
ss
co
n
t
r
ib
u
te
s
to t
h
e a
mp
lificatio
n
o
f
c
r
a
sh
r
i
s
k. T
h
i
s
m
i
g
h
t be beca
u
s
e t
h
at a
fi
rm
w
it
h
h
i
g
h
e
r
co
nn
ecte
dn
e
ss
h
a
s
a
g
r
eate
r
ca
p
acity a
nd
e
f
ficie
n
cy to
r
eceive a
nd
spr
ea
d
e
x
t
r
e
m
e
s
e
n
ti
m
e
n
t. T
h
e
g
e
n
e
r
al t
r
a
nsm
i
ss
io
n
o
f
t
h
e
s
i
g
n
al
pr
o
m
ote
s
f
o
rm
i
n
g
a
n
i
rr
atio
n
al
i
n
v
e
st
m
e
n
t at
m
os
ph
e
r
e
,
wh
ic
h
i
n
c
r
e
as
e
s i
nf
o
rm
atio
n
asy
mm
e
t
r
y a
nd
t
h
e
p
ossibility o
f
cov
-
e
r
i
n
g
“
ba
d
n
e
w
s
”
,
r
aisi
n
g
t
h
e
volatility o
f
stoc
k
pr
ic
e
s a
nd
c
r
as
h
pr
o
n
e
n
e
ss
[
65
]
.
O
ur
r
e
s
u
lts
a
l
s
o
ha
ve e
c
o
n
o
m
i
c
m
e
an
i
n
g
s
, t
ha
t i
s
,
a
chan
ge o
f
o
n
e
s
t
an
d
a
rd
d
evi
a
tio
n
i
n
Fi
r
m
S
e
n
tix
r
e
su
lt
s
i
n
a
n
a
ppr
o
x
i
m
ately 11.4
%
(1.2895 × 0.0518/0.5845) i
n
c
r
ea
s
e i
n
NCSKEW.
4.3. R
o
b
ust Test
4.3.1. E
n
d
o
g
e
n
eity
:
I
ns
t
rum
e
n
tal Va
r
iable
s
To eli
m
i
n
ate t
h
e
p
ote
n
tial e
ff
ect
s
o
f
e
n
d
o
g
e
n
eity o
n
o
ur
r
e
su
lt
s
, t
h
i
s
s
t
u
d
y
s
elect
s
t
h
e
i
n
st
rum
e
n
tal va
r
iabl
e
s
(
IVs
)
o
f
i
nd
e
p
e
nd
e
n
t va
r
iabl
e
s.
S
i
m
ila
r
to Hao a
nd
Xio
n
g
(
2021
)
[
66
]
,
w
e
s
ele
c
t t
h
e
a
ve
r
a
ge v
a
l
u
e o
f
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
to ot
h
e
r
fi
r
ms
i
n
t
h
e
i
d
e
n
tical
s
tate
(
Inst
r
ument_S
)
a
n
d
t
h
e ave
r
a
g
e val
u
e o
f
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t co
nn
ecte
d
n
e
ss
to
ot
h
e
r
fi
rms
i
n
t
h
e i
d
e
n
tical i
n
d
us
t
r
y
(
Inst
r
ument_I
)
, t
h
at i
s
,
Inst
r
ument_S =
∑
j
∈st
a
te
Fi
r
mSenti
x
j
/(n − 1),
(
19
)
Inst
r
ument_I =
∑
j
∈i
n
d
u
st
ry
Fi
r
mSenti
x
j
/(m − 1),
(
20
)
w
h
e
r
e
∑
st
a
te
Fi
r
mSenti
x
i
s
t
h
e
s
u
m
o
f
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
i
n
t
h
e
sam
e
s
t
a
te,
an
d
n i
s
t
h
e
n
u
mb
e
r
o
f
fi
r
ms
i
n
t
h
e
sam
e
s
t
a
te.
∑
i
n
d
u
st
ry
Fi
r
mSenti
x
i
s
t
h
e
s
u
m
o
f
s
e
n
ti
m
e
n
t
co
nn
e
ct
e
dn
e
ss
i
n
t
h
e
s
a
m
e
i
ndus
t
r
y, a
nd
m i
s
t
h
e
num
b
e
r
o
f
t
h
e
fi
rms
i
n
t
h
e
s
a
m
e
i
ndus
t
r
y.
F
ur
t
h
e
r
m
o
r
e,
s
i
nc
e t
h
e
r
e
ma
y
b
e
a
m
u
t
u
a
l
ca
u
sa
l
r
el
a
tio
nsh
i
p
b
et
w
ee
n
t
h
e
c
urr
e
n
t
s
to
c
k
p
r
i
c
e
cr
a
sh
r
i
s
k a
n
d
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
,
w
e a
dd
t
h
e
c
u
rr
e
n
t
s
to
c
k
p
r
i
c
e
cr
a
sh
r
i
s
k
a
s
a co
n
t
r
ol va
r
iabl
e
to t
h
e
fi
rs
t
-s
ta
g
e
r
e
g
r
e
ss
io
n
. To avoi
d
t
h
e
i
mp
act o
f
t
h
e
co
mp
a
n
y
’s
s
tock
pr
ic
e
c
r
a
sh
r
i
s
k
fr
o
m
t
h
e
pr
e
vio
us
y
e
a
r
o
n
t
h
e
c
urr
e
n
t
s
ta
g
e
,
w
e
a
dd
o
n
e
-
y
e
a
r-
la
gg
e
d
s
to
c
k
p
r
i
c
e
cr
a
sh
r
i
s
k to t
h
e
s
e
c
o
n
d
s
ta
g
e. T
h
e
r
e
s
u
lt
s
o
f
t
h
e t
w
o
-s
ta
g
e
r
e
g
r
e
ss
io
n
p
r
o
c
e
ss
a
r
e
sh
o
w
n
i
n
Table 3.
T
h
e
fi
r
st
r
e
g
r
e
ssio
n
sta
g
e
is s
h
o
wn
i
n
t
h
e
fi
r
st col
umn
o
f
Tabl
e
3.
S
e
n
ti
m
e
n
t co
nn
e
ct
e
d-
n
e
ss
i
s
p
o
s
itively
c
o
rr
elate
d
w
it
h
t
h
e
s
e t
w
o i
ns
t
r
u
m
e
n
tal va
r
ia
b
le
s
at t
h
e 1
%
s
i
g
n
ifi
c
a
nc
e
l
e
v
e
l. T
h
e
r
e
su
lt
s
o
f
t
h
e
s
e
co
nd
r
e
g
r
e
ss
io
n
s
ta
g
e
a
r
e
sh
o
wn
i
n
t
h
e
s
e
co
nd
col
umn
o
f
Tabl
e
3.
T
h
e i
n
ve
s
to
r
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
i
s
p
o
s
itively
c
o
rr
elate
d
w
it
h
s
to
c
k
pr
i
c
e
c
r
a
sh
r
i
s
k
(
β = 29.885
)
, a
n
d
t
h
e
s
i
g
n
ifica
n
ce level i
s
1
%
. T
h
e
r
e
f
o
r
e, t
h
e e
n
d
o
g
e
n
eity a
n
aly
s
i
s
supp
o
r
t
s
o
ur
m
ai
n
r
e
su
lt
s
.
Ta
b
le 3.
S
e
n
ti
m
e
n
t co
nn
ecte
d
n
e
ss
a
n
d
c
r
a
sh
r
i
s
k
:
i
ns
t
rum
e
n
t va
r
iable
m
et
h
o
d
.
Va
r
ia
b
le
NCSKEW
i,t−1
Col
u
m
n
(
1
)
Fi
r
mS
e
n
t
i
x
i,
t
Fi
r
mSenti
x
i,t
Col
u
m
n
(
2
)
NCSKE
W
i,
t
29.8850 ***
(
10.28
)
−0.0970 ***
(
−7.91
)
NCSKEW
i,t
Inst
r
umet_I
i,t
Inst
r
ument_S
i,t
DTURN
i,t
0.0021 **
(
2.35
)
1.1341 ***
(
7.79
)
0.1348 **
(
2.11
)
−0.0004
(
−1.27
)
0.0105 ***
(
4.30
)
Ent
r
o
p
y
2025, 27, 345
12 o
f
26
Ta
b
le 3. C
o
nt.
Va
r
ia
b
le
Col
u
m
n
(
1
)
Fi
r
mS
e
n
t
i
x
i,
t
Col
u
m
n
(
2
)
NCSKE
W
i,
t
R
O
E
i,t
MB
i,t
V
LE
i,t
SIZE
i,t
O
paq
ue
i,t
Co
ns
ta
n
t
−1.83 × 10
−6
*
(
−1.68
)
−0.0013
(
−1.27
)
−0.0018
(
−0.20
)
0.0013
(
0.71
)
0.0057
(
0.16
)
−0.0600
(
−0.84
)
6.54 × 10
−5
***
(
5.22
)
−0.3050 ***
(
−19.99
)
0.0072
(
0.04
)
−0.0184
(
−0.62
)
0.5941
(
1.17
)
−10.9749 ***
(
−9.48
)
Fi
r
m
Ye
a
r
Ye
s
Ye
s
Ye
s
Ye
s
F
A
d
j
.R
2
O
b
s
.
30.74
0.085
5936
109.82
0.161
5564
Not
e:
T
h
is tabl
e
s
h
o
w
s t
h
e
a
n
alysis
r
e
s
u
lts o
f
i
n
st
rum
e
n
tal va
r
iabl
e
s. Col
umn
(
1
)
is t
h
e
r
e
s
u
lt o
f
t
h
e
fi
r
st
r
e
g
r
e
ssio
n
sta
g
e
, a
nd
col
umn
(
2
)
is t
h
e
r
e
s
u
lt o
f
t
h
e
s
e
co
nd
r
e
g
r
e
ssio
n
sta
g
e
. T
h
e
t t
e
st val
u
e
s a
r
e
i
n
p
a
r
e
n
t
h
e
s
e
s,
wh
e
r
e
***, **, a
n
d
* i
n
d
icate
s
tati
s
tical
s
i
g
n
ifica
n
ce at t
h
e 1
%
, 5
%
a
n
d
10
%
level
s
,
r
e
sp
ectively.
4.3.2. E
n
d
o
g
e
n
eity
:
P
r
o
p
e
ns
ity
S
co
r
e
M
atc
h
i
n
g
Give
n
t
h
e i
nh
e
r
e
n
t
cha
lle
n
ge
s
i
n
q
u
an
ti
f
yi
n
g
an
d
c
o
n
t
r
olli
n
g
c
e
r
t
a
i
n
c
o
u
n
t
r
y
-
level
a
nd
fi
rm-
l
e
v
e
l
f
acto
rs
,
w
e
e
mp
loy t
h
e
pr
o
p
e
ns
ity
s
co
r
e
-m
atc
h
e
d
m
e
t
h
o
d
to
r
e
-
e
val
u
at
e
o
ur
m
o
d
e
l. T
h
is
m
e
t
h
o
d
allo
w
s
u
s to acco
un
t
f
o
r
t
h
e
d
i
ff
e
r
e
n
c
e
s i
n
fi
rm
c
h
a
r
act
e
r
istics b
e
t
w
ee
n
h
ig
h
an
d
lo
w
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
.
Sp
e
c
ifi
ca
lly,
w
e
c
l
ass
i
f
y fi
r
ms
w
it
h
s
e
n
ti
m
e
n
t
co
nn
ecte
d
n
e
ss
above t
h
e
m
e
d
ia
n
a
s
h
avi
n
g
h
i
g
h
s
e
n
ti
m
e
n
t co
nn
ecte
d
n
e
ss
, a
n
d
vice ve
rs
a.
I
n
t
h
e
fi
r
st sta
g
e
,
w
e
i
n
cl
ud
e
t
h
e
fi
rm-
l
e
v
e
l co
n
t
r
ol va
r
iabl
e
s o
f
h
i
g
h
s
e
n
ti
m
e
n
t co
nn
e
ct
e
dn
e
ss
a
s
o
u
tli
n
e
d
i
n
E
qu
atio
n
(
15
)
, a
n
d
calc
u
late t
h
e
pr
o
p
e
ns
ity
s
co
r
e
f
o
r
eac
h
s
a
mp
le. T
h
e
n
e
x
t
s
t
e
p
i
n
volv
e
s
m
atc
h
i
n
g
e
ac
h
fi
rm
y
e
a
r
i
n
t
h
e
h
i
g
h-s
e
n
ti
m
e
n
t
-
co
nn
e
ct
e
dn
e
ss
g
r
o
up
w
it
h
t
h
e
n
ea
r
e
s
t
pr
o
p
e
ns
ity
s
co
r
e i
n
t
h
e lo
w
-s
e
n
ti
m
e
n
t
-
co
nn
ecte
d
n
e
ss
g
r
o
up
.
P
a
n
el A
pr
e
s
e
n
t
s
t
h
e
r
e
su
lt
s
o
f
t
h
e fi
rs
t
-s
ta
g
e
r
e
g
r
e
ss
io
n
, a
nd
P
a
n
el
B
pr
ovi
d
e
s
t
h
e e
s
ti
m
atio
n
o
f
t
h
e
r
elatio
nsh
i
p
bet
w
ee
n
s
e
n
ti
m
e
n
t co
nn
ecte
d
n
e
ss
a
n
d
c
r
a
sh
r
i
s
k i
n
Table 4. Notably, t
h
e
s
e
r
e
su
lt
s
r
e
m
ai
n
un
c
h
a
n
g
e
d
w
h
e
n
co
mp
a
r
e
d
to o
ur
m
ai
n
r
e
g
r
e
ss
io
n
i
n
Table 2.
Ta
b
le 4.
P
r
o
p
e
ns
ity
s
co
r
e
m
atc
h
i
n
g
r
e
su
lt
s
.
P
a
n
el A
:
Fi
rs
t
-s
ta
g
e
pr
o
p
e
ns
ity
s
co
r
e
m
atc
h
i
n
g
R
O
E
i,t
MB
i,t
V
LE
i,t
SIZE
i,t
Hi
g
h
s
e
n
ti
m
e
n
t co
nn
ecte
d
n
e
ss
Va
r
iable
DTURN
i,t
0.0091 ***
(
5.32
)
−2.846 × 10
−5
(
−1.14
)
−0.1492 ***
(
−10.97
)
−0.0923
(
−1.47
)
−0.0411 ***
(
−3.73
)
Ent
r
o
p
y
2025, 27, 345
13 o
f
26
Ta
b
le 4. C
o
nt.
O
paq
ue
i,t
NCSKEW
i,t−1
C
o
nst
a
nt
Ye
a
r
O
b
s.
P
seu
d
o
R
2
0.0344
(
0.13
)
−0.0322 ***
(
−2.78
)
1.1358 ***
(
4.35
)
Ye
s
5564
0.003
P
a
n
el
B
:
S
e
n
ti
m
e
n
t co
nn
ecte
d
n
e
ss
a
n
d
c
r
a
sh
r
i
s
k
C
o
nt
r
o
ls
O
b
s.
A
d
j
.R
2
Hig
h
sentiment c
o
nnecte
d
ness
6.8832 ***
(
9.08
)
Ye
s
2878
0.155
Note
:
T
h
i
s
table
r
e
p
o
r
t
s
t
h
e
m
atc
h
i
n
g
m
et
h
o
d
o
f
s
e
n
ti
m
e
n
t co
nn
ecte
d
n
e
ss
a
n
d
c
r
a
sh
r
i
s
k
pr
o
p
e
ns
ity
s
co
r
e.
P
a
n
el A
i
s
t
h
e e
s
ti
m
atio
n
r
e
s
u
lt o
f
t
h
e
P
r
o
b
it
m
o
d
el i
n
t
h
e fi
rs
t
s
ta
g
e. T
h
e
d
e
p
e
n
d
e
n
t va
r
ia
b
le o
f
t
h
e fi
rs
t
s
ta
g
e i
s
h
i
g
h
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
. I
f
t
h
e
c
o
nn
e
c
te
d
n
e
ss
i
s
h
i
g
h
e
r
t
h
a
n
t
h
e
s
a
m
p
le
m
e
d
ia
n
, t
h
i
s
i
n
d
i
c
ato
r
va
r
ia
b
le i
s
o
n
e,
an
d
ot
h
e
rw
i
s
e, it i
s
ze
r
o.
W
e
r
eg
r
e
ss
e
d
h
ig
h
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
f
o
r
fi
r
m
f
e
a
t
ur
e
s
an
d
u
s
e t
h
e e
s
ti
ma
te
d
c
oe
f
fi
c
ie
n
t
s
f
r
o
m
t
h
e fi
r
s
t
-s
tage
m
o
d
el to
c
al
c
u
late t
h
e
pr
o
p
e
ns
ity
sc
o
r
e
f
o
r
ea
ch
o
bs
e
r
vatio
n
i
n
t
h
e
s
a
m
p
le.
W
e t
h
e
n
m
atc
h
e
d
e
ac
h
fi
rm
y
e
a
r
i
n
t
h
e
h
i
g
h-
co
nn
e
ct
e
dn
e
ss
g
r
o
up
w
it
h
t
h
e
fi
rm
y
e
a
r
i
n
t
h
e
lo
w-
co
nn
e
ct
e
dn
e
ss
g
r
o
up
a
n
d
obtai
n
e
d
t
h
e clo
s
e
s
t
pr
o
p
e
ns
ity
s
co
r
e.
P
a
n
el
B
r
e
p
o
r
t
s
t
h
e
O
L
S
r
e
su
lt
s
o
f
pr
o
p
e
ns
ity
s
co
r
e
m
atc
h
i
n
g
s
a
mp
le
s
to te
s
t t
h
e
i
n
te
rm
e
d
iate
r
elatio
nsh
i
p
bet
w
ee
n
s
e
n
ti
m
e
n
t co
nn
ecte
d
n
e
ss
a
n
d
c
r
a
sh
r
i
s
k. T
h
e val
u
e
s
o
f
t
h
e t te
s
t a
r
e i
n
p
a
r
e
n
t
h
e
s
e
s
,
w
it
h
*** i
n
d
icati
n
g
s
tati
s
tical
s
i
g
n
ifica
n
ce at t
h
e 1
%
.
4.3.3. T
h
e E
ff
ect o
f
Net
w
o
r
k Ce
n
t
r
ality
M
ea
sur
e
s
To
f
u
r
t
h
e
r
d
etect
r
ob
u
s
t
n
e
ss
o
f
t
h
e
r
e
s
u
lt
s
,
w
e al
s
o te
s
t t
h
e
r
elatio
nsh
i
p
bet
w
ee
n
co
n-
n
e
ct
e
dn
e
ss
a
nd
c
r
a
sh
r
i
s
k
us
i
n
g
ot
h
e
r
n
e
t
w
o
r
k co
nn
e
ctio
n
m
e
a
sur
e
s
.
S
i
n
c
e
t
h
e
i
mp
o
r
ta
n
c
e
o
f
t
h
e
s
e
n
ti
m
e
n
t s
p
illov
e
r
n
e
t
w
o
r
k
m
e
as
ur
e
s t
h
e
s
p
illov
e
r
o
f
a fi
rm
,
w
e
co
n
si
d
e
r
t
h
e
i
mp
act
o
f
L
e
a
d
e
Ra
n
k c
e
n
t
r
ality
(
LR
)
, K
-
co
r
e
n
e
ss
c
e
n
t
r
ality
(
KC
)
, a
nd
P
a
g
e
Ra
n
k c
e
n
t
r
ality
(
P
R
)
o
n
fi
rm
c
r
a
sh
r
i
s
k.
T
h
e
e
sti
m
at
e
d
r
e
s
u
lts a
r
e
ill
u
st
r
at
e
d
i
n
Tabl
e
5.
W
e
s
h
o
w
t
h
at t
h
e
r
e
latio
n
s
h
i
p
b
e
t
w
ee
n
s
e
n
ti
m
e
n
t
s
p
illove
r
n
et
w
o
r
k
c
e
n
t
r
a
lity i
n
d
exe
s
(
LR, KC
an
d
P
R
)
an
d
c
r
ash
r
i
s
k
r
e
ma
i
ns
p
o
s
itive,
sh
o
w
i
n
g
t
h
at a fi
rm
locate
d
i
n
a
m
o
r
e i
mp
o
r
ta
n
t
p
o
s
itio
n
h
a
s
a
h
i
g
h
e
r
c
r
a
sh
r
i
s
k.
T
h
i
s
r
e
su
lt co
n
fi
rms
t
h
e
r
ob
us
t
n
e
ss
o
f
o
ur
m
ai
n
r
e
g
r
e
ss
io
n
i
n
Tabl
e
2.
S
i
n
c
e
t
h
e
i
mp
o
r
ta
n
c
e
o
f
a
c
o
m
p
an
y i
n
t
h
e
s
e
n
ti
m
e
n
t
s
p
illove
r
n
et
w
o
r
k give
s
it
a
h
ig
h
ab
ility to
acq
u
i
r
e
an
d
pr
o
p
a
g
at
e
i
nf
o
rm
atio
n
i
n
a c
e
r
tai
n
g
r
o
up
o
r
d
i
ff
e
r
e
n
t
g
r
o
ups
, t
h
i
s
r
ai
s
e
s
t
h
e
h
e
rd
e
ff
e
ct a
nd
i
rr
atio
n
al be
h
avio
r
o
f
i
n
ve
s
to
rs
, ca
us
i
n
g
c
r
a
sh
r
i
s
k
[
29
]
.
Ta
b
le 5. Re
su
lt
s
w
it
h
alte
rn
ative
n
et
w
o
r
k ce
n
t
r
ality
m
ea
sur
e
s
.
Va
r
ia
b
le
Col
u
m
n
(
1
)
NCSKE
W
i,
t
Col
u
m
n
(
2
)
NCSKE
W
i,
t
Col
u
m
n
(
3
)
DU
V
OL
i,
t
Col
u
m
n
(
4
)
DU
V
OL
i,
t
LD
i,t
0.0735 **
(
2.48
)
KC
i,t
0.0253 ***
(
13.54
)
P
R
i,t
23.4394 **
(
2.17
)
DTURN
i,t
0.0031
(
1.39
)
0.0030
(
1.32
)
0.0137 ***
(
3.94
)
0.0030 **
(
2.32
)
Ent
r
o
p
y
2025, 27, 345
14 o
f
26
Ta
b
le 5. C
o
nt.
Va
r
ia
b
le
Col
u
m
n
(
1
)
NCSKE
W
i,
t
Col
u
m
n
(
2
)
NCSKE
W
i,
t
Col
u
m
n
(
3
)
DU
V
OL
i,
t
Col
u
m
n
(
4
)
DU
V
OL
i,
t
R
O
E
i,t
1.08 × 10
−5
(
1.33
)
1.23 × 10
−5
(
1.49
)
1.91 × 10
−5
*
(
1.76
)
1.20 × 10
−5
(
1.46
)
MB
i,t
−0.3556 ***
(
−26.93
)
−0.3546 ***
(
−26.86
)
−0.3541 ***
(
−30.39
)
−0.3547 ***
(
−26.86
)
V
LE
i,t
0.1128
(
0.72
)
0.1068
(
0.69
)
0.2083
(
1.33
)
0.1070
(
0.69
)
SIZE
i,t
0.0350
(
1.18
)
0.0351
(
1.17
)
−0.0032
(
−0.11
)
0.0319
(
1.17
)
O
paq
ue
i,t
0.6218
(
1.28
)
0.6153
(
1.26
)
0.8447 *
(
1.71
)
0.6911
(
1.42
)
NCSKEW
t−1
−0.1096 ***
(
−8.66
)
−0.1074 ***
(
−8.58
)
−0.1043 ***
(
−7.99
)
−0.1076 ***
(
−8.59
)
Co
ns
ta
n
t
0.0173
(
0.02
)
−0.0628
(
−0.09
)
0.0258
(
0.03
)
−1.0572
(
−1.49
)
Fi
rm
FE
Yea
r
FE
Ye
s
Ye
s
Ye
s
Ye
s
Ye
s
Ye
s
Ye
s
Ye
s
O
b
s
e
r
vatio
ns
A
d
j. R
2
F
-s
tat
5564
0.242
145.92
5564
0.243
142.55
5564
0.170
271.67
5564
0.243
142.80
Not
e:
T
h
i
s
tabl
e
pr
e
s
e
n
t
s
r
e
g
r
e
ss
io
n
r
e
su
lt
s
us
i
n
g
alt
e
rn
ativ
e
n
e
t
w
o
r
k
c
e
n
t
r
ality
m
e
a
sur
e
s
. Col
umns
(
2
)
–
(
4
)
r
e
p
o
r
t
r
e
su
lt
s
us
i
n
g
Lea
d
e
r
Ra
n
k, K
-
co
r
e
n
e
ss
, a
n
d
P
a
g
eRa
n
k ce
n
t
r
ality,
r
e
sp
ectively. All
sp
ecificatio
ns
i
n
cl
u
d
e fi
rm
a
n
d
yea
r
fi
x
e
d
e
ff
ect
s
. Rob
us
t t
-s
tati
s
tic
s
i
n
p
a
r
e
n
t
h
e
s
e
s
. ***
p
< 0.01, **
p
< 0.05, *
p
< 0.1.
4.3.4. T
h
e
P
ee
r
E
ff
ect
W
e
f
ur
t
h
e
r
ex
am
i
n
e t
h
e
p
ee
r
e
ff
e
c
t o
n
s
to
c
k
c
r
ash
r
i
s
k. Ext
an
t lite
r
a
t
ur
e
(
e.g.,
B
u
s
t
a-
man
te
an
d
F
r
é
sa
rd
(
2020
)
[
67
]
; C
h
e
n
et
a
l.
(
2019
)
[
68
])
has
d
o
c
u
m
e
n
te
d
t
ha
t fi
r
ms
o
f
te
n
m
i
m
ic
p
ee
rs’
d
eci
s
io
ns
ba
s
e
d
o
n
ob
s
e
r
vable c
h
a
r
acte
r
i
s
tic
s
,
u
n
d
e
rs
co
r
i
n
g
t
h
e
r
ole o
f
p
ee
r
i
n
fl
u
e
nc
e i
n
sha
p
i
n
g
s
to
c
k
pr
i
c
e
d
y
nam
i
cs
.
Sp
e
c
ifi
ca
lly,
M
u
ge
r
man
et
a
l.
(
2014
)
[
69
]
an
d
Ka
u
s
tia et al.
(
2015
)
[
70
]
h
i
g
h
li
g
h
t t
h
at fi
rms
m
ay i
n
fl
u
e
n
ce ot
h
e
rs
t
hr
o
ug
h
p
ee
r
e
ff
ect. To
i
s
olate t
h
e
s
ta
n
d
alo
n
e e
ff
ect o
f
s
e
n
ti
m
e
n
t co
nn
ecte
d
n
e
ss
fr
o
m
p
ee
r-
i
n
d
u
ce
d
c
r
a
sh
r
i
s
k,
w
e
i
n
co
rp
o
r
at
e
p
ee
r
co
mp
a
n
i
e
s
’
c
r
as
h
r
is
k
as a co
n
t
r
ol va
r
iabl
e
,
wh
ic
h
is
d
e
fi
n
e
d
as t
h
e
av
e
r
a
g
e
c
r
a
sh
r
i
s
k o
f
all fi
rms
i
n
t
h
e
s
a
m
e i
n
d
us
t
r
y
:
m − 1
P
ee
r
i
=
∑
j
∈i
n
d
u
st
ry
NCSKEW
j
,
(
21
)
w
h
e
r
e m
d
e
n
ote
s
t
h
e total
n
u
mb
e
r
o
f
fi
rms
i
n
t
h
e i
n
du
s
t
r
y. Col
u
mn
(
1
)
o
f
Ta
b
le 6
r
e
p
o
r
t
s
t
h
e
r
e
s
u
lt
s
a
f
te
r
co
n
t
r
olli
n
g
f
o
r
p
ee
r
c
r
a
sh
r
i
s
k. Nota
b
ly, Fi
r
mSenti
x
r
e
m
ai
ns
s
i
g
n
ifica
n
tly
n
e
g
ative i
n
pr
e
d
icti
n
g
c
r
a
sh
r
i
s
k,
r
ei
nf
o
r
ci
n
g
t
h
e
r
ob
us
t
n
e
ss
o
f
o
ur
ba
s
eli
n
e fi
n
d
i
n
g
s
.
Ta
b
le 6. T
h
e e
ff
ect o
f
s
e
n
ti
m
e
n
t
n
et
w
o
r
k
un
d
e
r
HFT a
n
d
p
ee
r
e
ff
ect.
Va
r
ia
b
le
Col
u
m
n
(
1
)
NCSKE
W
i,
t
Col
u
m
n
(
2
)
NCSKE
W
i,
t
Fi
r
mSenti
x
i,t
1.0501 **
(
2.04
)
1.3399 ***
(
2.93
)
P
ee
r
0.7710 *
(
−1.95
)
Ent
r
o
p
y
2025, 27, 345
15 o
f
26
Ta
b
le 6. C
o
nt.
Va
r
ia
b
le
Col
u
m
n
(
1
)
NCSKE
W
i,
t
Col
u
m
n
(
2
)
NCSKE
W
i,
t
HFT
2.4320 **
(
2.30
)
DTURN
i,t
0.0032 **
(
2.41
)
0.0040 **
(
2.37
)
R
O
E
i,t
1.21 × 10
−5
(
1.53
)
1.18 × 10
−5
(
1.53
)
MB
i,t
−0.3547 ***
(
−26.83
)
−0.2916 ***
(
−26.70
)
V
LE
i,t
0.1107
(
0.72
)
0.1168
(
0.70
)
SIZE
i,t
−0.0329 *
(
1.80
)
0.0220
(
1.03
)
NCSKEW
i,t−1
−0.6241 *
(
1.82
)
−0.2012 **
(
1.76
)
O
paq
ue
i,t
−1.1056 ***
(
−7.59
)
−1.1056 ***
(
−7.59
)
Co
ns
ta
n
t
−0.395
(
−0.56
)
−1.408
(
−0.26
)
Fi
rm
FE
Yea
r
FE
Ye
s
Ye
s
Ye
s
Ye
s
F
-s
tat
A
d
j. R
2
O
b
s
.
128.03
0.215
5564
142.80
0.245
5564
Note: T
h
i
s
t
ab
le
pr
e
s
e
n
t
s
r
e
s
u
lt
s
c
o
n
t
r
olli
n
g
f
o
r
HFT
ma
r
ket
s
ize
(
Col
u
mn
1
)
an
d
p
ee
r
e
ff
e
c
t
s
(
Col
u
mn
2
)
. All
sp
ecificatio
ns
i
n
cl
u
d
e fi
rm
fi
x
e
d
e
ff
ect
s
. Rob
us
t t
-s
tati
s
tic
s
i
n
p
a
r
e
n
t
h
e
s
e
s
. ***
p
< 0.01, **
p
< 0.05, *
p
< 0.1.
4.3.5. T
h
e E
ff
ect o
f
Hi
g
h-
F
r
e
qu
e
n
cy T
r
a
d
i
n
g
A
s
c
o
m
p
etitio
n
i
n
te
ns
ifie
s
,
h
i
g
h-fr
e
q
u
e
nc
y t
r
a
d
i
n
g
(
HFT
)
h
a
s
r
a
p
i
d
ly evolve
d
. HFT
u
s
e
s
c
o
m
pu
te
r
a
lgo
r
it
hms
to ex
p
loit
sh
o
r
t
-
te
r
m
pr
i
c
e
p
a
tte
r
ns
, i
n
volvi
n
g ext
r
e
m
ely
fas
t
o
rd
e
r
p
l
ac
e
m
e
n
t, exe
c
u
tio
n
,
an
d
canc
ell
a
tio
n
. It i
s
f
u
lly
a
u
to
ma
te
d
,
cha
r
ac
te
r
ize
d
b
y
ve
r
y
h
i
g
h-
vol
um
e t
r
a
d
i
n
g
a
n
d
sh
o
r
t
-
te
rm
h
ol
d
i
n
g
p
e
r
io
d
s
, ty
p
ically
w
it
h
i
n
m
ic
r
o
s
eco
n
d
s
.
HFT
h
a
s
ca
u
g
h
t t
h
e atte
n
tio
n
o
f
aca
d
e
m
ic
s
,
pr
actitio
n
e
rs
, a
n
d
m
a
r
ket
r
e
g
u
lato
rs
,
w
h
o a
r
e
i
nc
r
ea
s
i
n
g
ly
c
o
nc
e
r
n
e
d
a
b
o
u
t it
s
i
m
p
a
c
t o
n
m
a
r
ket
q
u
ality
[
71
]
.
M
o
r
eove
r
,
m
a
n
y
s
t
ud
ie
s
p
oi
n
t o
u
t t
h
at H
F
T
w
o
u
l
d
spr
ea
d
i
n
g
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t, alte
r
i
n
g
t
h
e t
r
a
d
i
n
g
d
y
n
a
m
ic
[
72
]
.
Z
h
a
n
g
(
2010
)
p
o
s
it
s
t
h
at HFT
h
a
s
p
e
rn
icio
us
e
ff
ect
s
o
n
U.
S
. ca
p
ital
m
a
r
ket
s
,
m
aki
n
g
s
tock
pr
i
c
e
s
ove
rr
e
ac
t to
f
u
n
d
am
e
n
t
a
l
n
e
w
s
[
73
]
.
Ja
rr
o
w
an
d
Pr
otte
r
(
2012
)
s
u
gge
s
t t
ha
t HFT
t
r
a
d
i
n
g
m
ay
dr
iv
e
m
a
r
k
e
t
pr
ic
e
s
fr
o
m
fund
a
m
e
n
tal val
u
e
s
a
nd
ca
us
e
g
r
e
at
e
r
volatility
[
74
]
.
I
n
t
h
i
s
s
ectio
n
,
w
e co
ns
i
d
e
r
t
h
e e
ff
ect o
f
HFT t
r
a
d
i
n
g
o
n
fi
rm
c
r
a
sh
r
i
s
k.
W
e co
ns
i
d
e
r
t
h
e
H
F
T
m
a
r
k
e
t siz
e
fr
o
m
2006 to 2021 as a co
n
t
r
ol va
r
iabl
e
(
t
h
e
H
F
T
m
a
r
k
e
t siz
e
d
ata
w
e
r
e
obtai
n
e
d
fr
o
m
h
tt
p
s
:
//
www
.i
b
i
s
w
o
r
l
d
.
c
o
m
/
u
n
ite
d
-s
tate
s
/
m
a
r
ket
-s
ize/
h
i
g
h-fr
e
q
u
e
nc
y
-
t
r
a
d
i
n
g
/4740/,
acc
e
ss
e
d
o
n
10
F
e
b
ru
a
r
y 2025
)
. Col
umn
(
2
)
o
f
Tabl
e
6 s
h
o
w
s t
h
e
r
e
latio
n
s
h
i
p
b
et
w
ee
n
Fi
r
mSenti
x
an
d
fi
r
m
c
r
ash
r
i
s
k. A
f
te
r
c
o
n
t
r
olli
n
g t
h
e HFT
ma
r
ket,
w
e
s
till fi
n
d
a
n
eg
a
tive i
m
p
ac
t o
f
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
o
n
c
r
ash
r
i
s
k. T
h
i
s
r
e
s
u
lt
s
u
gge
s
t
s
t
ha
t
w
h
ile HFT
m
ay i
m
p
a
c
t
m
a
r
ket t
r
a
d
i
n
g
a
n
d
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t, t
h
e
c
o
n
ta
g
io
n
o
f
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t
r
e
m
ai
ns
a key
f
acto
r
i
n
fl
u
e
n
ci
n
g
c
r
a
sh
r
i
s
k.
Ent
r
o
p
y
2025, 27, 345
16 o
f
26
4.4. C
o
m
pa
r
is
o
n
o
f
Fi
r
ms
’
In
d
i
v
i
d
u
a
l Sentiment
a
n
d
t
h
e Sentiment C
o
nnecte
d
ness
W
e co
mp
a
r
e t
h
e i
mp
act o
f
s
e
n
ti
m
e
n
t a
n
d
it
s
co
nn
ecte
d
n
e
ss
o
n
c
r
a
sh
r
i
s
k i
n
t
h
i
s
p
a
r
t.
Di
ff
e
r
e
n
t
fr
o
m
fi
rms’
i
nd
ivi
du
al
s
e
n
ti
m
e
n
t, co
nn
ecte
dn
e
ss
f
oc
us
e
s
o
n
t
h
e i
n
te
rd
e
p
e
nd
e
n
ce
a
n
d
co
-m
ove
m
e
n
t a
m
o
n
g
i
n
d
ivi
d
u
al
s
e
n
ti
m
e
n
t
s
,
w
h
ic
h
a
r
e
m
o
r
e likely to
h
a
pp
e
n
d
ur
i
n
g
ext
r
e
m
e
c
o
n
d
itio
ns
[
40,41
]
. T
h
e
r
e
f
o
r
e,
w
e
c
o
ns
i
d
e
r
t
h
e i
n
fl
u
e
nc
e o
f
s
e
n
ti
m
e
n
t
an
d
it
s
co
nn
ecte
d
n
e
ss
w
h
e
n
s
e
n
ti
m
e
n
t i
s
sur
g
i
n
g
o
r
d
ecli
n
i
n
g
.
W
e
d
ivi
d
e t
h
e
w
h
ole
sam
p
le i
n
to fi
r
s
t
an
d
l
as
t 50
%
q
u
an
tile
s
u
b-sam
p
le
s
. Co
n
t
r
a
r
y
to t
h
e
w
h
ole
sam
p
le, t
h
e
up
s
ur
ge
an
d
d
o
w
ns
ur
ge
s
e
n
ti
m
e
n
t
s
u
b-sam
p
le
s
sh
o
w
ext
r
e
m
e
c
o
n
d
itio
ns
o
f
fi
rm
s
e
n
ti
m
e
n
t, i
n
w
h
i
ch
t
h
e
c
o
nn
e
c
te
d
n
e
ss
i
s
m
o
r
e likely to
b
e
c
o
n
ta
g
io
u
s
.
W
e te
s
t t
h
e
s
e
n
ti
m
e
n
t a
n
d
it
s
co
nn
ecte
d
n
e
ss
i
n
t
h
e
w
h
ole
s
a
mp
le, t
h
e
upsur
g
e
su
b
-s
a
mp
le
an
d
t
h
e
d
o
w
ns
ur
ge
s
u
b-sam
p
le.
W
e fi
n
d
t
ha
t
b
ot
h
t
h
e fi
r
m’s
i
n
d
ivi
du
a
l
s
e
n
ti
m
e
n
t
an
d
t
h
e
s
e
n
ti
m
e
n
t
s
p
illove
r
n
et
w
o
r
k
a
r
e
s
ig
n
ifi
can
tly
r
el
a
te
d
to
c
r
ash
r
i
s
k i
n
w
h
ole
sam
p
le
s
.
Ho
w
eve
r
,
u
n
d
e
r
ext
r
e
m
e
ma
r
ket
c
o
n
d
itio
ns
, g
r
o
up
s
(
1
)
an
d
(
2
)
i
n
T
ab
le 7
d
e
m
o
ns
t
r
a
te
t
ha
t
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
has
a
s
ig
n
ifi
can
t i
m
p
ac
t o
n
c
r
ash
r
i
s
k.
M
o
r
e i
m
p
o
r
t
an
tly,
w
e fi
n
d
t
ha
t t
h
e i
n
fl
u
e
nc
e o
f
s
e
n
ti
m
e
n
t o
n
c
r
ash
r
i
s
k i
s
d
i
ff
e
r
e
n
t i
n
t
h
r
ee
s
u
b-sam
p
le
s
,
an
d
it i
s
s
ig
n
ifi
can
t o
n
ly
w
h
e
n
u
s
i
n
g
a
w
h
ole
sam
p
le,
w
h
i
ch
sh
o
w
s
t
ha
t
s
e
n
ti
m
e
n
t i
s
r
el
a
te
d
to
c
r
ash
r
i
s
k
u
n
d
e
r
n
o
r
ma
l
ma
r
ket
c
o
n
d
itio
ns
. It
r
efle
c
t
s
t
ha
t t
h
e i
n
fl
u
e
nc
e o
f
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t i
n
t
h
e
upsur
g
e a
n
d
d
o
w
nsur
g
e
s
ta
g
e
s
o
n
c
r
a
sh
r
i
s
k
m
ai
n
ly
s
te
ms
fr
o
m
s
e
n
ti
m
e
n
t co
nn
ecte
d
n
e
ss
.
T
h
e
s
e
r
e
s
u
lt
s
sh
o
w
t
ha
t
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
i
s
a
b
ette
r
pr
e
d
i
c
to
r
o
f
c
r
ash
r
i
s
k,
e
s
p
e
c
i
a
lly
dur
i
n
g
ma
r
ket ext
r
e
m
e
s
,
w
h
i
ch
h
ig
h
lig
h
t
s
t
h
e
n
ee
d
to i
nc
l
ud
e
s
e
n
ti
m
e
n
t
c
o
n-
n
e
c
tio
ns
i
n
r
e
s
e
a
r
ch
m
o
d
el
s
. D
ur
i
n
g
s
eve
r
e
ma
r
ket
s
t
r
e
ss
, i
n
ve
s
to
r
e
m
otio
ns
b
e
c
o
m
e
h
i
g
h
ly
d
ivi
d
e
d
, ca
us
i
n
g
i
n
ve
s
to
rs
to act
s
i
m
ila
r
ly. T
h
i
s
pr
oce
ss
qu
ickly
spr
ea
d
s
s
e
n
ti
m
e
n
t
s
h
oc
k
s ac
r
oss t
h
e
m
a
r
k
e
t
[
46,75
]
. As a
r
e
s
u
lt, t
h
e
e
ff
e
ct o
f
s
e
n
ti
m
e
n
t c
h
a
n
g
e
s o
n
stoc
k
pr
ic
e
s
b
e
co
m
e
s
w
e
a
k
e
r
. Ho
w
e
v
e
r
,
wh
e
n
i
n
v
e
sto
r
s
e
n
ti
m
e
n
t s
pr
e
a
d
s t
hr
o
u
g
h
co
mmun
icatio
n
c
h
a
n-
n
el
s
o
r
m
e
d
ia
r
e
p
o
r
t
s
, it ca
n
i
nf
ect ot
h
e
r
i
n
ve
s
to
rs
, i
n
c
r
ea
s
i
n
g
m
a
r
ket
s
e
n
ti
m
e
n
t volatility
a
n
d
m
aki
n
g
s
tock
pr
ice
s
w
i
n
g
s
bi
gg
e
r
[
76
]
.
Ta
b
le 7. Co
mp
a
r
i
s
o
n
o
f
s
e
n
ti
m
e
n
t co
nn
ecte
d
n
e
ss
a
n
d
s
e
n
ti
m
e
n
t.
Va
r
ia
b
le
G
r
o
up
(
1
)
U
p
s
u
r
g
e
G
r
o
up
(
2
)
Do
wn
s
u
r
g
e
G
r
o
up
(
3
)
W
h
ole
NCSKE
W
i,
t
NCSKE
W
i,
t
NCSKE
W
i,
t
NCSKE
W
i,
t
NCSKE
W
i,
t
NCSKE
W
i,
t
Fi
r
mSenti
x
i,t
1.6873 **
(
2.55
)
1.3522 *
(
1.75
)
1.2895 ***
(
2.82
)
Sentiment
i,t
DTURN
i,t
R
O
E
i,t
MB
i,t
V
LE
i,t
SIZE
i,t
O
paq
ue
i,t
NCSKEW
i,t−1
C
o
nst
a
nt
0.0042
(
1.50
)
2.77 × 10
−5
***
(
4.44
)
−0.4242 ***
(
−17.70
)
0.1311
(
0.61
)
0.0513
(
1.52
)
0.3163
(
0.46
)
−0.0913 ***
(
−4.77
)
−0.8977
(
−1.05
)
−0.0044
(
−0.08
)
0.0042
(
−1.65
)
2.53 × 10
−5
***
(
−3.8
)
−0.4253 ***
(
−17.69
)
0.1215
(
−0.56
)
0.0526
(
−1.57
)
0.3162
(
−0.46
)
−0.0924 ***
(
−4.78
)
−0.3054
(
−0.37
)
0.0090
(
1.13
)
−0.0008
(
−0.32
)
−0.3307 ***
(
−16.15
)
0.5017
(
1.35
)
0.1533
(
1.61
)
1.5577
(
1.47
)
−0.1539 ***
(
−5.31
)
−4.1146 *
(
−1.86
)
0.0253
(
−0.17
)
−0.0039
(
−1.05
)
8.11 × 10
−6
(
−0.74
)
−0.3251 ***
(
−20.49
)
0.0596
(
−0.27
)
−0.0401
(
−0.41
)
1.0433
(
−1.55
)
−0.1203 ***
(
−7.47
)
1.6847
(
−0.76
)
0.0032
(
1.39
)
1.21 × 10
−5
(
1.53
)
−0.3547 ***
(
−26.83
)
0.1107
(
0.72
)
0.0328
(
1.11
)
0.6249
(
1.27
)
−0.1073 ***
(
−8.58
)
−0.3993 ***
(
−0.56
)
−0.1064 **
(
−2.47
)
0.0012
(
−0.46
)
3.14 × 10
−6
(
0.27
)
−0.3559 ***
(
−26.94
)
0.1241
(
−0.8
)
0.0282
(
−0.94
)
0.6331
(
−1.29
)
−0.1073 ***
(
−8.50
)
0.2521
(
−0.35
)
Ent
r
o
p
y
2025, 27, 345
17 o
f
26
Ta
b
le 7. C
o
nt.
Va
r
ia
b
le
G
r
o
up
(
1
)
U
p
s
u
r
g
e
G
r
o
up
(
2
)
Do
wn
s
u
r
g
e
G
r
o
up
(
3
)
W
h
ole
NCSKE
W
i,
t
NCSKE
W
i,
t
NCSKE
W
i,
t
NCSKE
W
i,
t
NCSKE
W
i,
t
NCSKE
W
i,
t
Fi
r
m Ye
s
Yea
r
Ye
s
Ye
s
Ye
s
Ye
s
Ye
s
Ye
s
Ye
s
Ye
s
Ye
s
Ye
s
Ye
s
F65.67
A
d
j
.R
2
0.232
O
b
s
.2760
69.0235.7496.33
0.40210.2430.257
276027902790
141.38146.01
0.2430.244
55645564
Note
:
T
h
i
s
ta
b
le i
s
t
h
e
r
e
s
u
lt o
f
b
ot
h
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
a
n
d
s
e
n
ti
m
e
n
t o
n
cr
a
sh
r
i
s
k. Hig
h
(
L
o
w
)
d
e
n
ote
s
t
h
at
s
e
n
ti
m
e
n
t is
h
i
g
h
(
lo
w)
;
50
%
sa
mp
l
e
d
e
n
ot
e
s t
h
at t
h
e
fi
r
st 50
%
o
f
t
h
e
sa
mp
l
e
is classifi
e
d
as t
h
e
h
i
g
h-
s
e
n
ti
m
e
n
t
s
t
a
ge,
an
d
t
h
e
s
e
c
o
n
d
50
%
i
s
c
l
ass
ifie
d
as
t
h
e lo
w
-s
e
n
ti
m
e
n
t
s
t
a
ge. A
s
a
w
h
ole, W
h
o
le
d
e
n
ote
s
t
h
e e
ff
e
c
t o
f
t
h
e
s
e
n
ti
m
e
n
t
o
r
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
o
n
s
to
c
k
pr
i
c
e. T
h
e t te
s
t v
a
l
u
e
s
a
r
e i
n
p
a
r
e
n
t
h
e
s
e
s
,
w
h
e
r
e ***, **,
an
d
* i
n
d
icate
s
tati
s
tical
s
i
g
n
ifica
n
ce at t
h
e 1
%
, 5
%
, a
n
d
10
%
level
s
,
r
e
sp
ectively.
4.5.
P
o
tenti
a
l C
ha
nnel An
a
l
y
sis
S
o
f
a
r
, o
ur
fi
nd
i
n
g
s s
h
o
w
a s
u
bsta
n
tial
p
ositiv
e
co
rr
e
latio
n
b
e
t
w
ee
n
s
e
n
ti
m
e
n
t co
nn
e
ct
-
e
dn
e
ss a
nd
t
h
e
li
k
e
li
h
oo
d
o
f
a stoc
k
m
a
r
k
e
t
m
e
lt
d
o
wn
. I
n
pur
s
u
it o
f
a
d
ee
p
e
r
und
e
r
sta
nd
i
n
g
o
f
t
h
e
u
n
d
e
r
lyi
n
g
p
ote
n
ti
a
l
chann
el t
ha
t li
n
k
s
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
an
d
s
to
c
k
pr
i
c
e
c
r
a
sh
r
i
s
k, a
m
o
d
e
r
ati
n
g
e
ff
ect
m
o
d
el i
s
e
s
tabli
sh
e
d
a
s
f
ollo
w
s
:
i
NCSKEW
i,t
=
α
+ β
1
Fi
r
mSenti
x
i,t
+ β
2
Fi
r
mSenti
x
i,t
∗
M
o
d
i,t
+ β
3
DTURN
i,t
+ β
4
MB
i,t
+ β
5
R
O
E
i,t
+ β
6
SIZE
i,t
+ β
7
LEV
,t
+ β
8
M
o
d
i,t
,
(
22
)
+ β
9
NCSKEW
i,t
+ β
10
O
paq
ue
i,t
+
θ
t
+ µ
i
+
ε
i,t
w
h
e
r
e
M
o
d
i,t
r
e
p
r
e
s
e
n
t t
h
e t
w
o
m
e
d
iato
rs
o
f
s
y
n
c
hr
o
n
icity
(
S
y
nc
h
i,t
)
a
n
d
acco
u
n
ti
n
g
co
n-
s
e
r
vatis
m
(
Csc
o
r
e
i,t
)
. T
h
e
va
r
iabl
e
S
y
nc
h
i,t
i
nd
icat
e
s t
h
e
co
-m
ov
e
m
e
n
t o
f
fi
rm
s a
nd
m
a
r
k
e
ts,
w
h
i
ch
i
s
u
s
e
d
as
a
pr
oxy
f
o
r
t
h
e
q
u
an
tity o
f
fi
r
m-s
p
e
c
ifi
c
i
nf
o
r
ma
tio
n
c
o
n
t
a
i
n
e
d
i
n
s
to
c
k
pr
i
c
e
f
o
r
fi
r
m
i i
n
yea
r
t, a
n
d
t
h
e va
r
ia
b
le Csc
o
r
e
i,t
i
n
d
i
c
ate
s
t
h
e a
cc
o
u
n
ti
n
g
c
o
ns
e
r
vati
sm
f
o
r
fi
rm
i i
n
yea
r
t,
w
h
ic
h
us
e
s
t
h
e i
n
d
e
x
d
evelo
p
e
d
by K
h
a
n
a
n
d
W
att
s
(
2009
)
[
77
]
.
4.5.1.
M
o
d
e
r
atio
n
E
ff
ect
s
o
f
S
tock
P
r
ice
S
y
n
c
hr
o
n
icity
Bu
il
d
i
n
g
o
n
t
h
e
t
h
e
o
r
e
tical
fr
a
m
e
w
o
r
k
, o
ur
a
n
alysis
r
e
v
e
als t
h
at fi
rm
s
w
it
h
h
i
g
h
e
r
s
e
n-
ti
m
e
n
t co
nn
ecte
d
n
e
ss
ca
n
r
a
p
i
d
ly acce
ss
m
o
r
e i
mp
act
fu
l
s
e
n
ti
m
e
n
t i
nf
o
rm
atio
n
co
mp
a
r
e
d
to t
h
e
i
r
p
ee
rs
.
Sp
e
cifically,
hu
b co
mp
a
n
i
e
s
m
iti
g
at
e
s
tock
pr
ic
e
c
r
a
sh
r
i
s
k by
r
e
du
ci
n
g
s
tock
p
r
ice
s
y
n
c
hr
o
n
icity, e
nh
a
n
ci
n
g
i
nf
o
rm
atio
n
t
r
a
ns
p
a
r
e
n
cy, a
n
d
c
u
r
bi
n
g
i
n
ve
s
to
r
i
rr
atio
n
al
-
ity
dur
i
n
g ext
r
e
m
e
s
e
n
ti
m
e
n
t
p
e
r
io
d
s
[
44
]
. T
h
e
s
e
m
e
chan
i
sms
m
otiv
a
te t
h
e i
nc
l
u
s
io
n
o
f
s
tock
pr
ice
s
y
n
c
hr
o
n
icity a
s
a
m
e
d
iati
n
g
va
r
iable i
n
o
ur
a
n
aly
s
i
s
. T
h
e e
mp
i
r
ical
r
e
su
lt
s
a
r
e
pr
e
s
e
n
te
d
i
n
Table 8.
To
un
cov
e
r
t
h
e
m
o
d
e
r
ati
n
g
i
mp
act o
f
stoc
k
pr
ic
e
sy
n
c
hr
o
n
is
m
,
w
e
pr
ovi
d
e
a
n
i
n
t
e
r
act
-
i
n
g
va
r
iable o
f
t
h
e
s
e
n
ti
m
e
n
t co
nn
ecte
d
n
e
ss
a
n
d
s
tock
pr
ice
s
y
n
c
hr
o
n
icity
(
Fi
r
mSenti
x
i,t
∗
S
y
nc
i,t
)
. Col
u
mn
(
2
)
r
eveal
s
t
h
at
s
to
c
k
r
et
u
rn
s
y
nchr
o
n
i
c
ity a
m
p
lifie
s
t
h
e i
n
fl
u
e
nc
e o
f
t
h
e
s
e
n
ti
m
e
n
t co
nn
ecte
dn
e
ss
o
n
s
tock
pr
ice c
r
a
sh
r
i
s
k. T
h
i
s
m
ay be
du
e to t
h
e
f
act t
h
at co
mp
a
-
n
i
e
s
w
it
h
h
i
g
h
stoc
k
pr
ic
e
sy
n
c
hr
o
n
izatio
n
m
ov
e
i
n
loc
k
st
e
p
w
it
h
t
h
e
m
a
r
k
e
t,
wh
ic
h
r
e
du
c
e
s
t
h
e t
r
a
nsp
a
r
e
n
cy o
f
fi
rm-sp
ecificity i
nf
o
rm
atio
n
a
n
d
r
ai
s
e
s
t
h
e bli
n
d
n
e
ss
i
n
i
n
ve
s
ti
n
g
, a
n
d
r
ai
s
e
s
t
h
e i
n
fl
u
e
n
ce o
f
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t.
Ent
r
o
p
y
2025, 27, 345
18 o
f
26
Ta
b
le 8. T
h
e e
ff
ect o
f
s
e
n
ti
m
e
n
t
n
et
w
o
r
k o
n
c
r
a
sh
r
i
s
k
un
d
e
r
d
i
ff
e
r
e
n
t co
n
d
itio
ns
.
Va
r
ia
b
le
Col
u
m
n
(
1
)
NCSKE
W
i,
t
Col
u
m
n
(
2
)
NCSKE
W
i,
t
Col
u
m
n
(
3
)
NCSKE
W
i,
t
S
y
nc
i,t
Fi
r
mSenti
x
i,t
× S
y
nc
i,t
Fi
r
mSenti
x
i,t
1.3399 ***
(
2.93
)
1.1584 **
(
2.49
)
−0.1346 **
(
−2.57
)
0.4429 **
(
2.05
)
Csc
o
r
e
i,t
Fi
r
mSenti
x
i,t
× Csc
o
r
e
i,t
DTURN
i,t
R
O
E
i,t
MB
i,t
i
LEV
,t
SIZE
i,t
NCSKEW
i,t−1
O
paq
ue
i,t
C
o
nst
a
nt
0.0031
(
1.38
)
1.08 × 10
−5
(
1.33
)
−0.3556 ***
(
−26.93
)
0.1128
(
0.72
)
0.0350
(
1.18
)
−0.1096 ***
(
−8.66
)
0.6218
(
1.28
)
0.0173
(
0.02
)
0.0046
(
1.60
)
1.18 × 10
−5
*
(
2.00
)
−0.3426 ***
(
−27.61
)
0.2268
(
1.49
)
0.1130 ***
(
2.70
)
−0.1036 ***
(
−8.79
)
0.7911 *
(
1.64
)
−2.4345 **
(
−2.47
)
0.3603 *
(
1.80
)
−0.1777 *
(
−1.94
)
0.0064 **
(
2.33
)
7.45 × 10
−6
(
1.11
)
−0.3510 ***
(
−26.61
)
0.1268
(
0.81
)
0.0922 **
(
2.33
)
−0.1076 ***
(
−8.46
)
0.7309
(
1.49
)
−1.8162 *
(
−1.93
)
Fi
rm
FE
Yea
r
FE
Ye
s
Ye
s
Ye
s
Ye
s
Ye
s
Ye
s
F
-s
tat
A
d
j. R
2
O
b
s
.
145.92
0.242
5564
140.23
0.272
5564
132.01
0.260
5564
Note: T
h
i
s
t
ab
le
r
e
p
o
r
t
s
t
h
e
m
o
d
e
r
a
ti
n
g e
ff
e
c
t o
f
s
to
c
k
s
y
nch
r
o
n
i
c
ity
an
d
acc
o
u
n
ti
n
g
c
o
ns
e
r
v
a
ti
sm
o
n
t
h
e
r
el
a
tio
nsh
i
p
b
et
w
ee
n
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
an
d
c
r
ash
r
i
s
k. Col
u
mn
(
1
)
ex
c
l
ud
e
s
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
;
Col
u
mn
(
2
)
i
nc
l
ud
e
s
s
to
c
k
s
y
nchr
o
n
i
c
ity i
n
te
r
a
c
tio
n
;
Col
u
mn
(
3
)
i
nc
l
ud
e
s
a
cc
o
u
n
ti
n
g
c
o
ns
e
r
vati
sm
i
n
te
r
a
c
tio
n
. t
-
s
tati
s
tic
s
i
n
p
a
r
e
n
t
h
e
s
e
s
. ***, **, *
d
e
n
ote
s
i
g
n
ifica
n
ce at t
h
e 1
%
, 5
%
, a
n
d
10
%
level
s
.
4.5.2.
M
o
d
e
r
atio
n
E
ff
ect
s
o
f
Acco
un
ti
n
g
Co
ns
e
r
vati
sm
(
C
s
co
r
e
)
A
s
exe
c
u
tive
s
m
ay ove
rs
tate
p
e
rf
o
rm
a
nc
e o
r
c
o
nc
eal
b
a
d
n
e
w
s
f
o
r
c
o
m
p
e
ns
atio
n
o
r
ca
r
ee
r
i
n
ce
n
tive
s
[
49
]
, fi
rms
acce
ss
ible to e
x
t
r
e
m
e
s
e
n
ti
m
e
n
t
s
oo
n
e
r
f
ace
g
ove
rn
a
n
ce
d
i
s
a
d
-
va
n
ta
g
e
s
du
r
i
n
g
p
e
r
io
d
s
o
f
h
ei
g
h
te
n
e
d
e
m
otio
n
. Acco
r
d
i
n
g
to
[
60,78,79
]
, t
h
e
s
up
e
r
vi
s
io
n
o
f
i
rr
atio
n
al be
h
avio
r
by acco
u
n
ti
n
g
p
r
o
x
y ca
n
r
e
du
ce t
h
e co
n
ceal
m
e
n
t o
f
“ba
d
n
e
w
s
” by
c
o
m
p
an
y
mana
ge
r
s
, t
h
e
r
e
b
y
r
e
du
c
i
n
g t
h
e
s
to
c
k
pr
i
c
e
r
i
s
k o
f
a
c
o
m
p
an
y
ca
u
s
e
d
b
y t
h
e
e
x
te
rn
alitie
s
e
ff
ect o
f
i
n
ve
s
to
rs
f
o
r
e
x
t
r
e
m
e
s
e
n
ti
m
e
n
t.
W
e
s
eek to
d
ete
r
m
i
n
e
w
h
et
h
e
r
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
w
it
h
a
h
ig
h
C
sc
o
r
e i
n-
cr
ea
s
e
s
t
h
e e
ff
e
c
t o
f
i
nf
o
rm
atio
n
a
s
y
mm
et
r
y o
n
t
h
e
r
i
s
k o
f
p
r
i
c
e
c
olla
p
s
e
s
[
77
]
. T
h
e
r
e
f
o
r
e,
w
e
pr
ovi
d
e
a
i
n
te
r
ac
ti
n
g v
a
r
i
ab
le o
f
t
h
e
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
v
a
r
i
ab
le
an
d
C
sc
o
r
e
(
Fi
r
mSenti
x
i,t
∗ Csc
o
r
e
i,t
)
. A
s
ill
u
s
t
r
a
te
d
i
n
T
ab
le 8, Col
u
mn
(
3
)
pr
e
s
e
n
t
s
t
h
e
r
e
s
u
lt
s
w
h
e
n
s
e
n
ti
m
e
n
t co
nn
ecte
d
n
e
ss
i
n
te
r
act
s
w
it
h
acco
un
ti
n
g
co
ns
e
r
vati
sm
(
Csc
o
r
e
)
, a
n
d
t
h
e
s
i
g
n
ifi
-
ca
n
t
n
e
g
ative coe
f
ficie
n
t o
f
Fi
r
mSenti
x
i,t
∗ Csc
o
r
e
i,t
i
n
d
icate
s
t
h
at acco
un
ti
n
g
co
ns
e
r
vati
sm
ca
n
m
iti
g
at
e
t
h
e
e
ff
e
ct o
f
t
h
e
s
e
n
ti
m
e
n
t co
nn
e
ct
e
dn
e
ss
o
n
s
tock
pr
ic
e
c
r
a
sh
r
i
s
k. T
h
i
s
co
u
l
d
b
e
b
e
ca
u
s
e
fi
rm
s
w
it
h
h
i
g
h
acco
un
ti
n
g
co
n
s
e
r
vatis
m
h
av
e
b
e
tt
e
r
ability to co
n
t
r
ol
e
x
c
e
ssiv
e
Ent
r
o
p
y
2025, 27, 345
19 o
f
26
mana
ge
r
i
a
l
s
el
f-
i
n
te
r
e
s
t, i
m
pr
ove i
nf
o
r
ma
tio
n
s
y
mm
et
r
y,
an
d
lo
w
e
r
illogi
ca
l i
n
ve
s
t
m
e
n
t
r
i
s
k. T
h
e
r
e
f
o
r
e, t
h
e acco
un
ti
n
g
co
ns
e
r
vati
sm
r
e
d
u
ce
s
t
h
e i
mp
act o
f
i
rr
atio
n
al
s
e
n
ti
m
e
n
t.
To ex
am
i
n
e
w
h
et
h
e
r
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
ex
ac
e
r
ba
te
s
t
h
e e
ff
e
c
t o
f
i
nf
o
r
ma
tio
n
as
y
mm
et
r
y o
n
c
r
ash
r
i
s
k
[
77
]
,
w
e i
n
t
r
o
du
c
e
an
i
n
te
r
ac
tio
n
te
r
m
b
et
w
ee
n
s
e
n
ti
m
e
n
t
c
o
n-
n
e
c
te
d
n
e
ss
(
Fi
r
mSenti
x
i,t
)
an
d
acc
o
u
n
ti
n
g
c
o
ns
e
r
v
a
ti
sm
(
Csc
o
r
e
i,t
)
. A
s
sh
o
w
n
i
n
T
ab
le 8,
Col
u
mn
(
3
)
r
e
p
o
r
t
s
t
h
e
r
e
s
u
lt
s
f
o
r
t
h
i
s
i
n
te
r
ac
tio
n
m
o
d
el. T
h
e
s
ig
n
ifi
can
tly
n
eg
a
tive
c
o
-
e
f
fi
c
ie
n
t o
n
Fi
r
mSenti
x
i,t
∗ Csc
o
r
e
i,t
i
n
d
i
ca
te
s
t
ha
t
acc
o
u
n
ti
n
g
c
o
ns
e
r
v
a
ti
sm
m
itig
a
te
s
t
h
e
i
mp
act o
f
s
e
n
ti
m
e
n
t co
nn
e
ct
e
dn
e
ss
o
n
s
tock
pr
ic
e
c
r
a
sh
r
i
s
k. T
h
i
s
fi
nd
i
n
g
ca
n
b
e
att
r
ib
u
t
e
d
to fi
rms
w
it
h
s
t
r
o
n
g
acco
un
ti
n
g
co
ns
e
r
vati
sm
h
avi
n
g
bette
r
ca
p
acity to co
ns
t
r
ai
n
m
a
n
a
g
e
-
r
i
a
l o
pp
o
r
t
u
n
i
sm
, e
nhanc
e i
nf
o
r
ma
tio
n
t
r
ans
p
a
r
e
nc
y,
an
d
r
e
du
c
e i
rr
a
tio
na
l i
n
ve
s
t
m
e
n
t
b
e
h
avio
r
.
B
y i
mpr
ovi
n
g
i
nf
o
rm
atio
n
q
u
ality, acco
un
ti
n
g
co
ns
e
r
vati
sm
t
h
e
r
e
by
w
e
a
k
e
ns
t
h
e
t
r
a
nsm
i
ss
io
n
o
f
i
rr
atio
n
al
s
e
n
ti
m
e
n
t ac
r
o
ss
m
a
r
ket
p
a
r
tici
p
a
n
t
s
.
4.6. Fu
r
t
h
e
r
An
a
l
y
ses
To
f
ur
t
h
e
r
ex
p
lo
r
e t
h
e
p
o
s
itive
r
el
a
tio
nsh
i
p
b
et
w
ee
n
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
an
d
s
to
c
k
pr
i
c
e
c
r
ash
r
i
s
k,
w
e
c
o
n
du
c
t
s
u
b-sam
p
le
r
eg
r
e
ss
io
ns
to ex
am
i
n
e
h
o
w
o
w
n
e
r
sh
i
p
s
t
ru
ct
ur
e a
n
d
fi
n
a
n
cial co
n
d
itio
n
m
o
d
e
r
ate t
h
i
s
a
ss
ociatio
n
a
f
te
r
co
n
t
r
olli
n
g
f
o
r
co
rp
o
r
ate
g
ove
rn
a
n
ce va
r
iable
s
.
4.6.1. T
h
e E
ff
ect o
f
Sh
a
r
e
h
ol
d
i
n
g
S
t
ru
ct
ur
e
W
e fi
rs
t e
x
a
m
i
n
e fi
rms’
sh
a
r
e
h
ol
d
i
n
g
s
t
ru
ct
ur
e
s
a
s
t
h
ey
f
o
rm
t
h
e
f
o
un
d
atio
n
o
f
co
rp
o
-
r
a
te gove
r
nanc
e.
Pr
io
r
r
e
s
e
a
r
ch
has
i
n
d
i
ca
te
d
t
ha
t o
w
n
e
r
sh
i
p
c
o
nc
e
n
t
r
a
tio
n
ma
y
h
el
p
to
all
e
viat
e
a
g
e
n
cy co
n
flict
s
b
e
t
w
ee
n
m
ajo
r
a
nd
m
i
n
o
r
ity
sh
a
r
e
h
ol
d
e
rs
by
e
nh
a
n
ci
n
g
i
nf
o
rm
a
-
tio
n
al asy
mm
e
t
r
y
w
it
h
i
n
fi
rm
s
[
79
]
. T
h
e
r
e
f
o
r
e
,
w
e
d
e
fi
n
e
a
h
i
g
h
o
wn
e
r
s
h
i
p
co
n
c
e
n
t
r
atio
n
i
f
a
fi
rm’s
i
n
itial
sh
a
r
e
h
ol
d
e
r
r
atio i
s
g
r
e
at
e
r
t
h
a
n
25
%
, a
nd
ot
h
e
r
fi
rms
h
av
e
a lo
w
sh
a
r
e
h
ol
d-
i
n
g
co
n
ce
n
t
r
atio
n
. T
h
e
r
e
su
lt
s
a
r
e
pr
e
s
e
n
te
d
i
n
Table 9.
Col
u
mns
(
1
)
an
d
(
2
)
i
n
T
ab
le 9
sh
o
w
t
ha
t
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
an
d
c
r
ash
r
i
s
k
a
r
e
p
o
s
itively
r
el
a
te
d
a
t t
h
e 1
%
s
ig
n
ifi
canc
e level i
n
c
o
m
p
an
ie
s
w
it
h
lo
w
sha
r
e
h
ol
d
i
n
g
c
o
nc
e
n
t
r
a
tio
ns
,
w
h
ile
n
o
r
el
a
tio
nsh
i
p
i
s
o
bs
e
r
ve
d
f
o
r
fi
r
ms
w
it
h
h
ig
h
sha
r
e
h
ol
d
i
n
g
d
i
s-
p
e
r
s
io
n
. T
h
i
s
i
n
d
i
ca
te
s
t
ha
t
a
lo
w
-c
o
nc
e
n
t
r
a
te
d
sha
r
e
h
ol
d
i
n
g
s
t
ru
c
t
ur
e
s
t
r
e
n
gt
h
e
ns
t
h
e
i
n
fl
u
e
nc
e o
f
s
p
illove
r
n
et
w
o
r
k
s
o
n
fi
r
m
c
r
ash
r
i
s
k. T
h
i
s
can
b
e ex
p
l
a
i
n
e
d
b
y t
h
e
fac
t t
ha
t a
lo
w
-
co
n
ce
n
t
r
ate
d
s
t
r
u
ct
u
r
e i
s
n
ot co
n
du
cive to
d
i
ss
e
m
i
n
ati
n
g
fi
rm-sp
ecific i
nf
o
rm
atio
n
,
wh
ic
h
co
u
l
d
i
n
c
r
e
as
e
i
n
v
e
sto
r
bli
ndn
e
ss a
nd
t
h
e
pr
obability o
f
i
rr
atio
n
al b
e
h
avio
r
[
50
]
.
Su
c
h
i
rr
a
tio
na
l
s
e
n
ti
m
e
n
t
ma
y t
h
e
n
ca
u
s
e
dr
ama
ti
c
s
w
i
n
g
s
i
n
s
to
c
k
pr
i
c
e
s
, t
h
e
r
e
b
y i
nc
r
e
as
i
n
g
c
r
a
sh
r
i
s
k.
Ta
b
le 9. Hete
r
o
g
e
n
eity a
n
aly
s
i
s
.
Va
r
ia
b
le
S
h
a
r
e
h
ol
d
i
ng
Co
n
ce
n
t
r
atio
n
Fi
n
a
n
cial Co
nd
itio
n
(
1
)
Hi
gh
(
2
)
Lo
w
(
3
)
No
n
-
Di
s
t
r
e
ss
e
d
(
4
)
Di
s
t
r
e
ss
DTURN
i,t
R
O
E
i,t
MB
i,t
V
LE
i,t
NCSKE
W
i,
t
NCSKE
W
i,
t
NCSKE
W
i,
t
NCSKE
W
i,
t
Fi
r
mSenti
x
i,t
3.7140
(
1.57
)
0.0014
(
0.38
)
−0.0002
(
−0.11
)
−0.4375 ***
(
−6.75
)
−0.9383 **
(
−2.20
)
1.1481 **
(
2.46
)
0.0098 ***
(
4.82
)
1.06 × 10
−5
(
1.34
)
−0.3497 ***
(
−26.03
)
0.1748
(
1.08
)
1.4280
(
1.64
)
0.0036
(
1.01
)
3.96 × 10
−5
(
1.30
)
−0.4078 ***
(
−14.56
)
0.2227
(
0.97
)
1.2249 **
(
2.25
)
0.0064 *
(
1.93
)
9.18 × 10
−6
(
1.16
)
−0.3416 ***
(
−23.59
)
0.0371
(
0.16
)
Ent
r
o
p
y
2025, 27, 345
20 o
f
26
Ta
b
le 9. C
o
nt.
Va
r
ia
b
le
S
h
a
r
e
h
ol
d
i
ng
Co
n
ce
n
t
r
atio
n
Fi
n
a
n
cial Co
nd
itio
n
(
1
)
Hi
gh
(
2
)
Lo
w
(
3
)
No
n
-
Di
s
t
r
e
ss
e
d
(
4
)
Di
s
t
r
e
ss
O
paq
ue
i,t
NCSKEW
i,t−1
C
o
nst
a
nt
NCSKE
W
i,
t
NCSKE
W
i,
t
NCSKE
W
i,
t
NCSKE
W
i,
t
SIZE
i,t
0.7077
(
1.54
)
−0.8956
(
−0.41
)
−0.1807 ***
(
−3.15
)
4.7837
(
0.93
)
0.0763 ***
(
2.64
)
0.6353
(
1.26
)
−0.1013 ***
(
−7.93
)
−1.490 **
(
−2.08
)
0.1063
(
0.80
)
0.9240
(
0.91
)
−0.1582 ***
(
−7.03
)
−2.3645
(
−0.74
)
0.0912 **
(
2.20
)
0.5222
(
0.93
)
−0.0903 ***
(
−6.18
)
−1.6573
(
−1.69
)
Fi
r
m
Ye
a
r
Ye
s
Ye
s
Ye
s
Ye
s
Ye
s
Ye
s
Ye
s
Ye
s
F
A
d
j
.R
2
O
b
s
.
168.90
0.217
285
144.95
0.211
5265
138.91
0.211
1215
115.72
0.246
4335
Note: T
h
i
s
t
ab
le
r
e
p
o
r
t
s
t
h
e
r
el
a
tio
nsh
i
p
b
et
w
ee
n
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
an
d
c
r
ash
r
i
s
k
u
n
d
e
r
d
i
ff
e
r
e
n
t
s
u
b-
sa
mp
l
e
s. Col
umn
s
(
1
)
a
nd
(
2
)
r
e
p
o
r
t
r
e
g
r
e
ssio
n
r
e
s
u
lts
f
o
r
co
n
c
e
n
t
r
at
e
d
a
nd
d
is
p
e
r
s
e
d
e
q
u
ity sa
mp
l
e
s. Col
umns
(
3
)
a
n
d
(
4
)
r
e
p
o
r
t
r
e
g
r
e
ss
io
n
r
e
su
lt
s
f
o
r
fi
n
a
n
cially
s
o
un
d
a
n
d
uns
table
s
a
mp
le
s
. Yea
r
_D
umm
y
r
e
pr
e
s
e
n
t
s
a
d
umm
y
va
r
iable abo
u
t ti
m
e. T
h
e t te
s
t val
u
e
s
a
r
e i
n
p
a
r
e
n
t
h
e
s
e
s
,
w
h
e
r
e ***, **, a
n
d
* i
n
d
icate
s
tati
s
tical
s
i
g
n
ifica
n
ce at t
h
e
1
%
, 5
%
, a
n
d
10
%
level
s
,
r
e
sp
ectively.
4.6.2. T
h
e E
ff
ect o
f
Fi
n
a
n
cial Co
n
d
itio
n
T
h
e fi
rm’s
fi
n
a
n
cial co
n
d
itio
n
i
s
o
ur
s
eco
n
d
m
ea
sur
e o
f
co
rp
o
r
ate
g
ove
rn
a
n
ce.
Wh
e
n
a
fi
r
m
i
s
i
n
fi
nanc
i
a
l t
r
o
u
b
le, i.e.,
cash
flo
w
i
s
i
ns
u
f
fi
c
ie
n
t, it
s
e
a
r
n
i
n
g
s
m
ig
h
t
n
ot
m
eet it
s
i
n
v
e
sto
r
s
’
e
x
p
e
ctatio
n
s,
wh
ic
h
w
o
u
l
d
r
e
s
u
lt i
n
a
d
e
cli
n
e
i
n
its stoc
k
pr
ic
e
a
nd
fi
rm
val
u
e
[
80
]
.
B
a
s
e
d
o
n
Za
n
g
(
2012
)
[
81
]
,
w
e c
h
oo
s
e −2.675 a
s
t
h
e bo
un
d
a
r
y o
f
fi
n
a
n
cial
d
i
s
t
r
e
ss
, t
h
at i
s
,
above −2.675 i
n
d
icate
s
fi
n
a
n
cial
d
i
s
t
r
e
ss
a
n
d
belo
w
−2.675 i
n
d
icate
s
fi
n
a
n
cial
s
tea
d
i
n
e
ss
.
W
e
sh
o
w
t
h
e
r
e
su
lt
s
i
n
Table 9. A
s
ill
us
t
r
ate
d
i
n
Table 9, Col
umns
(
3
)
a
n
d
(
4
)
su
gg
e
s
t
t
h
at t
h
e a
ss
ociatio
n
bet
w
ee
n
s
e
n
ti
m
e
n
t co
nn
ecte
d
n
e
ss
a
n
d
c
r
a
sh
r
i
s
k i
s
m
o
r
e
pr
o
m
i
n
e
n
t i
n
fi
n
a
n
cial
-
d
i
s
t
r
e
ss
fi
rms
b
u
t
n
ot i
n
fi
n
a
n
cially
s
o
u
n
d
fi
rms
. T
h
i
s
r
e
s
u
lt
m
ea
ns
t
h
at a
s
o
u
n
d
fi
n
a
n
cial co
n
d
itio
n
d
ec
r
ea
s
e
s
t
h
e e
ff
ect o
f
sp
illove
r
n
et
w
o
r
k
s
o
n
fi
rm
c
r
a
sh
r
i
s
k. T
h
i
s
m
ay
b
e
b
e
ca
u
s
e i
n
ve
s
to
r
s
a
r
e likely to
p
a
y
m
o
r
e
a
tte
n
tio
n
to t
h
e
b
e
ha
vio
r
o
f
t
h
e exe
c
u
tive
s
o
f
d
i
s
t
r
e
ss
e
d
c
o
m
p
an
ie
s
. T
h
e
mana
ge
m
e
n
t o
f
d
i
s
t
r
e
ss
e
d
fi
r
ms
i
s
i
nc
li
n
e
d
to
man
i
pu
l
a
te
e
a
r
n
i
n
g
s
mana
ge
m
e
n
t to
c
ove
r
up
t
h
e
r
e
a
l
s
it
u
a
tio
n
o
f
t
h
e fi
r
m’s
f
u
n
d
am
e
n
t
a
l
s
w
h
e
n
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t
b
e
c
o
m
e
s
ext
r
e
m
e
[
82
]
. T
h
i
s
i
nc
r
e
as
e
s
t
h
e o
p
aq
u
e i
n
s
u
ch
(
d
i
s
t
r
e
ss
e
d
)
fi
rms
,
w
h
ic
h
w
o
u
l
d
i
n
c
r
ea
s
e
s
t
h
e c
r
a
sh
pr
o
n
e
n
e
ss
.
5. Co
n
cl
u
s
io
n
s
T
h
i
s
p
a
p
e
r
e
x
a
m
i
n
e
s
h
o
w
s
e
n
ti
m
e
n
t co
nn
e
ct
e
dn
e
ss
b
e
t
w
ee
n
fi
rms
i
mp
act
s
s
toc
k
pr
ic
e
c
r
ash
r
i
s
k.
W
e
a
r
g
u
e t
ha
t
c
o
m
p
an
ie
s
w
it
h
h
ig
h
e
r
c
o
nn
e
c
te
d
n
e
ss
ha
ve
a
h
ig
h
e
r
ab
ility to
r
e
c
e
iv
e
a
nd
d
iss
e
m
i
n
at
e
s
e
n
ti
m
e
n
t,
e
x
ac
e
r
bati
n
g
i
n
v
e
sto
r’
s bli
ndn
e
ss a
nd
i
rr
atio
n
ality. T
h
is
ca
n
s
e
v
e
r
e
ly
pr
o
p
a
g
at
e
b
e
h
avio
rs
lik
e
“
h
e
rd
i
n
g
e
ff
e
ct
s
”
a
nd
“
e
n
t
hus
ia
sm
”
,
u
lti
m
at
e
ly l
e
a
d
to
pr
ic
e
c
r
as
h
e
s. Co
n
t
r
olli
n
g
f
o
r
ot
h
e
r
d
e
t
e
rm
i
n
a
n
ts, s
e
n
ti
m
e
n
t co
nn
e
ct
e
dn
e
ss i
n
c
r
e
m
e
n
tally
f
o
r
e
cas
t
s
c
r
ash
r
i
s
k
ab
ove ot
h
e
r
pr
e
d
i
c
to
r
s
. F
ur
t
h
e
r
m
o
r
e,
s
e
n
ti
m
e
n
t i
n
fl
u
e
nc
e
s
o
n
c
r
ash
r
i
s
k
dur
i
n
g
up
t
urns
a
nd
d
o
wn
t
urns
a
pp
e
a
r
dr
iv
e
n
by co
nn
e
ct
e
dn
e
ss
b
e
t
w
ee
n
fi
rms
. T
h
e
s
e
fi
nd
i
n
g
s
pr
ov
e
r
ob
u
st to
m
o
d
e
ls i
n
co
rp
o
r
ati
n
g
sy
n
c
hr
o
n
icity, acco
un
ti
n
g
co
n
s
e
r
vatis
m
, a
nd
gove
r
nanc
e
q
u
a
lity. T
w
o
-s
t
a
ge le
as
t
sq
u
a
r
e
s
e
s
ti
ma
tio
n
an
d
pr
o
p
e
ns
ity
sc
o
r
e
ma
t
ch
i
n
g
a
ddr
e
ss
e
nd
o
g
e
n
eity co
n
ce
rns
.
W
e fi
nd
t
h
at
s
tock
s
y
n
c
hr
o
n
icity a
mp
lifie
s
co
nn
ecte
dn
e
ss’s
Ent
r
o
p
y
2025, 27, 345
21 o
f
26
e
ff
e
c
t
s
o
n
c
r
ash
r
i
s
k,
w
h
ile
c
o
ns
e
r
v
a
ti
sm
m
itig
a
te
s
t
h
e e
ff
e
c
t. Fi
na
lly,
w
e
ana
lyze
h
o
w
o
w
n
e
r
sh
i
p
s
t
ru
c
t
ur
e a
n
d
fi
n
a
nc
ial
d
i
s
t
r
e
ss
m
o
d
e
r
ate t
h
e
r
elatio
nsh
i
p
b
et
w
ee
n
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
an
d
c
r
ash
r
i
s
k. T
h
i
s
r
e
s
e
a
r
ch
sh
e
d
s
lig
h
t o
n
c
o
n
d
itio
ns
i
n
fl
u
e
nc
i
n
g
h
o
w
s
e
n
ti
m
e
n
t
d
i
sp
e
rs
e
s
bet
w
ee
n
fi
rms
a
n
d
m
ate
r
ialize
s
i
n
s
tock
pr
ice
s
.
W
e
c
o
n
t
r
i
b
u
te to
u
n
d
e
r
s
t
an
d
i
n
g t
h
e
r
ole o
f
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
i
n
c
o
rp
o
r
a
te
g
ov
e
rn
a
n
c
e
a
nd
fi
n
a
n
cial
r
i
s
k.
B
y
e
x
a
m
i
n
i
n
g
t
h
e
r
e
latio
nsh
i
p
b
e
t
w
ee
n
fi
rm-
l
e
v
e
l
s
e
n
ti
m
e
n
t
co
n
ta
g
io
n
, acco
un
ti
n
g
pr
actic
e
s, a
nd
stoc
k
r
e
t
urn
s,
w
e
e
x
t
e
nd
pr
io
r
lit
e
r
at
ur
e
by i
d
e
n
ti
f
yi
n
g
h
o
w
s
e
n
ti
m
e
n
t co
nn
ectivity i
n
fl
u
e
n
ce
s
s
tock
p
r
ice c
r
a
sh
r
i
s
k.
Ou
r
fi
n
d
i
n
g
s
h
i
g
h
li
g
h
t t
h
at
s
e
n
ti
m
e
n
t
s
p
illove
r
n
et
w
o
r
k
s
p
l
a
y
a
c
r
iti
ca
l
r
ole i
n
pr
o
p
a
g
a
ti
n
g
c
r
ash
r
i
s
k,
s
u
gge
s
ti
n
g
t
ha
t t
h
e
s
e
s
y
s
te
m
i
c
li
n
k
a
ge
s
sh
o
u
l
d
b
e i
nc
o
rp
o
r
a
te
d
i
n
to
r
i
s
k
mana
ge
m
e
n
t
f
r
am
e
w
o
r
k
s
.
O
ve
r
all, t
h
i
s
r
e
s
ea
r
c
h
d
ee
p
e
ns
o
ur
un
d
e
rs
ta
n
d
i
n
g
o
f
h
o
w
sh
a
r
e
d
i
n
ve
s
to
r
ps
yc
h
olo
g
y a
n
d
i
nf
o
r
ma
tio
n
d
i
ff
u
s
io
n
t
h
r
o
u
g
h
fi
nanc
i
a
l
n
et
w
o
r
k
s
i
m
p
ac
t
ma
r
ket o
u
t
c
o
m
e
s
. To
m
itig
a
te
c
r
ash
r
i
s
k, fi
r
ms
an
d
r
eg
u
l
a
to
r
s
sh
o
u
l
d
pr
io
r
itize
m
o
n
ito
r
i
n
g
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
alo
n
g
s
i
d
e t
r
a
d
itio
n
al
r
i
s
k
m
ea
sur
e
s
.
Ou
r
s
t
ud
y co
mb
i
n
e
s
s
ocial
n
et
w
o
r
k
s
w
it
h
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t,
p
r
e
s
e
n
ti
n
g
a
n
i
nn
ova
-
tive be
h
avio
r
al fi
n
a
n
ce
p
e
rsp
ective. T
h
eo
r
etically, it
p
io
n
ee
rs
a
n
e
w
a
ppr
oac
h
to fi
n
a
n
cial
ma
r
ket
ana
ly
s
i
s
b
y
f
o
c
u
s
i
n
g o
n
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t i
n
te
r
ac
tio
n
. Fi
r
s
t, i
n
ve
s
to
r
i
n
te
r
ac-
tio
n
n
et
w
o
r
k
s
h
el
p
u
s
u
n
d
e
r
s
t
an
d
t
h
e
r
el
a
tio
nsh
i
p
b
et
w
ee
n
i
nf
o
r
ma
tio
n
d
i
ss
e
m
i
na
tio
n
an
d
ma
r
ket
r
e
ac
tio
ns
.
Wh
ile
b
e
ha
vio
r
a
l fi
nanc
e
f
o
c
u
s
e
s
o
n
h
o
w
i
n
ve
s
to
r
s
pr
o
c
e
ss
i
n-
f
o
r
ma
tio
n
an
d
ma
ke
d
e
c
i
s
io
ns
,
pr
io
r
s
t
ud
ie
s
sh
o
w
t
ha
t
ma
r
ket
s
e
n
ti
m
e
n
t
an
d
i
n
ve
s
to
r
be
h
avio
r
a
r
e o
f
te
n
i
n
fl
u
e
n
ce
d
by e
x
te
rn
al eve
n
t
s
a
nd
collective e
m
otio
ns
, a
s
d
e
m
o
ns
t
r
ate
d
by E
d
m
a
ns
et al.
(
2007
)
[
83
]
a
nd
Ab
ud
y
e
t al.
(
2022
)
[
84
]
.
S
ocial
n
e
t
w
o
r
k
s
a
m
o
n
g
i
n
v
e
s
to
rs
p
r
ovi
d
e a
f
a
s
t
ch
a
nn
el
f
o
r
i
nf
o
rm
atio
n
s
p
r
ea
d
. I
n
ve
s
to
r
i
n
te
r
a
c
tio
n
n
et
w
o
r
k
s
e
n
a
b
le
pu
b-
li
c
i
nf
o
r
ma
tio
n
to
b
e
q
u
i
c
kly
d
i
ss
e
m
i
na
te
d
am
o
n
g i
n
ve
s
to
r
s
,
s
ig
n
ifi
can
tly e
nhanc
i
n
g t
h
e
i
m
p
ac
t o
f
i
nf
o
r
ma
tio
n
r
ele
as
e
(
Cook
s
o
n
et
a
l., 2024
[
85
])
. T
h
e
r
e
f
o
r
e, i
n
teg
r
a
ti
n
g
s
o
c
i
a
l
n
et
w
o
r
k
s
i
n
to
b
e
ha
vio
r
a
l fi
nanc
e
a
llo
w
s
f
o
r
a
m
o
r
e i
n-
d
e
p
t
h
s
t
ud
y o
f
ma
r
ket
r
e
s
p
o
ns
e
s
to i
nf
o
rm
atio
n
.
S
eco
n
d
, co
m
bi
n
i
n
g
i
n
ve
s
to
r
m
ic
r
o
-
be
h
avio
r
al c
h
a
r
acte
r
i
s
tic
s
w
it
h
n
et
w
o
r
k
s
t
ru
ct
ur
e
s
h
el
ps
us
un
d
e
rs
ta
n
d
belie
f
d
ive
r
g
e
n
ce a
n
d
t
r
a
d
i
n
g
d
y
n
a
m
ic
s
. I
n
s
ocial
n
et
w
o
r
k
s
,
d
i
ff
e
r
e
n
t
i
n
ve
s
to
r
s’
i
n
te
rpr
et
a
tio
ns
an
d
b
elie
f
d
i
ff
e
r
e
nc
e
s
r
eg
a
rd
i
n
g t
h
e
sam
e i
nf
o
r
ma
tio
n
t
r
igge
r
t
r
a
d
i
n
g
b
e
h
avio
r
(P
e
n
g
a
nd
Z
h
a
n
g
, 2024
[
52
])
.
U
si
n
g
social
n
e
t
w
o
r
k
d
ata,
w
e
ca
n
st
ud
y t
h
e
e
x
t
e
n
t o
f
b
e
li
e
f
d
iv
e
r
g
e
n
c
e
a
nd
its i
mp
act o
n
t
r
a
d
i
n
g
vol
um
e
a
nd
m
a
r
k
e
t liq
u
i
d
ity. T
hr
o
u
g
h
i
n
t
e
r
actio
n
s i
n
social
n
e
t
w
o
r
k
s, i
n
v
e
sto
r
s
e
n
g
a
g
e
i
n
social l
e
a
rn
i
n
g
a
nd
co
n
ti
nu
o
u
sly
upd
at
e
t
h
e
i
r
b
e
li
e
f
s. T
h
is l
e
a
rn
i
n
g
pr
oc
e
ss is i
n
fl
u
e
n
c
e
d
by
f
acto
r
s s
u
c
h
as social
n
e
t
w
o
r
k
st
ru
ct
ur
e
,
i
nf
o
rm
atio
n
qu
ality, a
n
d
i
n
d
ivi
d
u
al co
g
n
itive bia
s
e
s
.
La
s
tly,
s
ocial
n
et
w
o
r
k
s
m
ay
spr
ea
d
m
i
s
i
nf
o
rm
atio
n
o
r
t
r
i
gg
e
r
ove
rr
eactio
ns
, ca
u
s
i
n
g
ma
r
ket
pr
i
c
e
s
to
d
evi
a
te
f
r
o
m
t
h
ei
r
t
ru
e v
a
l
u
e,
w
h
i
ch
i
s
c
ru
c
i
a
l
f
o
r
b
e
ha
vio
r
a
l fi
nanc
e
r
e
s
e
a
r
ch
. I
n
pr
ac
ti
c
e,
c
o
mb
i
n
i
n
g
b
e
ha
vio
r
a
l fi
nanc
e
w
it
h
s
o
c
i
a
l
n
et
w
o
r
k
ana
ly
s
i
s
o
ff
e
r
s
r
e
g
u
lato
r
y bo
d
ie
s
e
nh
a
n
ce
d
m
a
r
ket
m
o
n
ito
r
i
n
g
tool
s
.
B
y e
x
a
m
i
n
i
n
g
i
nf
o
rm
atio
n
flo
w
a
n
d
i
n
v
e
s
to
r
s
e
n
ti
m
e
n
t o
n
s
ocial
n
e
t
w
o
r
k
s
,
r
e
g
u
lato
rs
ca
n
pr
o
mp
tly
d
e
t
e
ct ill
e
g
al activiti
e
s
lik
e
m
a
r
ket
m
a
n
i
pu
latio
n
a
n
d
m
i
s
i
nf
o
rm
atio
n
, e
nsur
i
n
g
m
a
r
ket o
r
d
e
r
.
M
o
r
eove
r
, t
h
i
s
r
e
s
ea
r
c
h
ai
d
s i
n
c
r
a
f
ti
n
g
so
und
p
olici
e
s to
pr
o
m
ot
e
t
h
e
h
e
alt
h
y
u
s
e
o
f
social
n
e
t
w
o
r
k
s i
n
fi
n
a
n
c
e
a
nd
pr
otect i
n
ve
s
to
r
i
n
te
r
e
s
t
s
.
T
h
i
s
s
t
ud
y o
ff
e
r
s
c
ru
c
i
a
l
a
d
vi
c
e
f
o
r
i
n
ve
s
to
r
s
. Fi
r
s
t, fi
r
ms
w
it
h
h
ig
h
s
e
n
ti
m
e
n
t
c
o
n-
n
e
c
te
d
n
e
ss
a
r
e
m
o
r
e
pr
o
n
e to
s
to
c
k
pr
i
c
e
c
r
ash
e
s
. Ret
a
il i
n
ve
s
to
r
s
can
u
s
e
s
e
n
ti
m
e
n
t
c
o
nn
e
c
te
d
n
e
ss
m
et
r
i
cs
(
e.g., Fi
r
mSenti
x
)
to
s
p
ot
c
e
n
t
r
a
lly
p
o
s
itio
n
e
d
c
o
m
p
an
ie
s
i
n
s
e
n-
ti
m
e
n
t
n
et
w
o
r
k
s
,
w
h
i
ch
ma
y ex
p
e
r
ie
nc
e
sha
rp
pr
i
c
e
s
w
i
n
g
s
dur
i
n
g
ma
r
ket
m
oo
d
sh
i
f
t
s
.
O
ur
r
e
s
ea
r
ch
al
s
o
h
i
g
h
li
g
h
t
s
t
h
e
n
ee
d
f
o
r
i
n
ve
s
to
r
s
to
b
e
m
i
n
d
f
u
l o
f
i
n
te
r
-
fi
r
m
s
e
n
ti
m
e
n
t
sp
illove
rs
. T
h
i
s
i
mp
lie
s
t
h
at
s
electi
n
g
un
co
nn
ecte
d
s
tock
s
f
o
r
s
ecto
r-
ba
s
e
d
a
ss
et allocatio
n
can
r
e
du
c
e
s
e
n
ti
m
e
n
t
-
li
n
ke
d
r
i
s
k
s
. I
n
sh
o
r
t, o
ur
s
t
ud
y
sh
o
w
s
t
ha
t
ma
r
ket
s
e
n
ti
m
e
n
t
c
o
n-
Ent
r
o
p
y
2025, 27, 345
22 o
f
26
t
a
gio
n
can
t
r
igge
r
c
r
ash
r
i
s
k
s
. T
h
e
r
e
f
o
r
e, i
n
ve
s
to
r
s
m
ig
h
t
n
ee
d
to
u
s
e te
chn
i
ca
l tool
s
to
m
o
n
ito
r
m
a
r
ket
s
e
n
ti
m
e
n
t o
r
a
dd
lo
w
-
co
rr
elate
d
a
ss
et
s
to t
h
ei
r
p
o
r
t
f
olio
s
.
A
u
t
h
or
C
o
n
t
r
i
bu
ti
o
ns
:
Co
nc
e
p
t
u
a
liz
a
tio
n
,
J
.C., G.H.
an
d
Y.
J
.
;
m
et
h
o
d
ology,
J
.C.
;
s
o
f
t
w
a
r
e,
J
.C.
;
v
a
li
d
a
tio
n
, Y.
J
.
;
f
o
r
ma
l
ana
ly
s
i
s
,
J
.C.
an
d
G.H.
;
i
n
ve
s
tig
a
tio
n
,
J
.C.
an
d
Y.
J
.
;
r
e
s
o
ur
c
e
s
,
J
.C.
;
d
a
t
a
c
u
r
atio
n
, J.C. a
n
d
Y.J.
;
w
r
iti
n
g
—
o
r
i
g
i
n
al
d
r
a
f
t
p
r
e
p
a
r
atio
n
, G.H.
;
w
r
iti
n
g
—r
evie
w
a
n
d
e
d
iti
n
g
, J.C.
;
vi
su
alizatio
n
, Y.J.
;
sup
e
r
vi
s
io
n
, J.C. a
n
d
G.H.
;
pr
oject a
d
m
i
n
i
s
t
r
atio
n
, J.C. All a
u
t
h
o
rs
h
ave
r
ea
d
a
n
d
a
g
r
ee
d
to t
h
e
pu
bli
sh
e
d
ve
rs
io
n
o
f
t
h
e
m
a
nus
c
r
i
p
t.
F
und
i
n
g
:
T
h
is
r
e
s
e
a
r
c
h
w
as
fund
e
d
by
t
h
e
t
h
e
C
h
i
n
a
P
os
t
d
oc
t
o
r
al
S
ci
e
n
c
e
F
o
und
a
t
io
n
(
No.
2024
M
762372
)
, t
h
e
Pr
oje
c
t
Supp
o
r
te
d
b
y
S
c
ie
n
tifi
c
Re
s
e
a
r
ch
F
u
n
d
o
f
H
u
nan
Pr
ovi
nc
i
a
l E
du
ca
tio
n
De
p
a
r
t
m
e
n
t
(
No. 24C0146
)
, a
nd
t
h
e
Labo
r
a
t
o
r
y o
f
Co
mpu
t
a
t
io
n
a
nd
A
n
aly
t
ic
s
o
f
Co
mp
l
e
x
M
a
n
a
g
e
m
e
n
t
S
y
s
te
ms
O
p
e
n
Fo
un
d
atio
n
(
No. 2024
S
KHX046
)
.
I
n
s
tit
u
tio
n
al Revie
w
Boa
r
d
State
m
e
n
t
:
Not a
pp
licable.
Data Availa
b
ility State
m
e
n
t
:
T
h
e
r
a
w
d
ata
s
upp
o
r
ti
n
g
t
h
e
c
o
nc
l
u
s
io
ns
o
f
t
h
i
s
a
r
ti
c
le
w
ill
b
e
m
a
d
e
available by t
h
e a
u
t
h
o
rs
o
n
r
e
qu
e
s
t.
Co
n
flict
s
of I
n
te
r
e
s
t
:
T
h
e a
u
t
h
o
rs
d
ecla
r
e
n
o co
n
flict
s
o
f
i
n
te
r
e
s
t.
A
pp
e
nd
i
x
A. Va
r
ia
b
le
M
ea
s
u
r
e
m
e
n
t
Defi
n
itio
n
DUV
O
L
Zsc
o
r
e
Csc
o
r
e
S
y
nc
h
SIZE
LEV
R
O
E
MB
O
paq
ue
DTURN
HFT
Va
r
ia
b
le
NCSKEW
T
h
e
n
e
g
ative coe
f
ficie
n
t o
f
s
ke
w
n
e
ss
, calc
u
late
d
by taki
n
g
t
h
e
n
e
g
ative o
f
t
h
e t
h
i
r
d
m
o
m
e
n
t o
f
fi
rm-
sp
ecific
w
eekly
r
et
urns
f
o
r
eac
h
yea
r
a
n
d
d
ivi
d
i
n
g
it by t
h
e c
u
bic
s
ta
n
d
a
r
d
d
eviatio
n
o
f
fi
rm-sp
ecific
w
eekly
r
et
urns
.
S
ee E
qu
atio
n
(
7
)
.
T
h
e ot
h
e
r
m
e
as
ur
e
m
e
n
t o
f
c
r
ash
r
i
s
k;
w
e
s
e
p
a
r
a
te
a
ll t
h
e
w
eek
s
w
it
h
fi
r
m-s
p
e
c
ifi
c
w
eekly
r
et
ur
ns
a
b
ove t
h
e a
nn
u
al ave
r
a
g
e
(
up
w
eek
s)
fr
o
m
t
h
o
s
e
w
it
h
fi
rm-s
p
e
c
ifi
c
r
et
u
rns
b
elo
w
t
h
e a
nn
u
al
m
ea
n
(
d
o
w
n
w
eek
s)
a
n
d
co
mpu
te t
h
e
s
ta
n
d
a
r
d
d
eviatio
n
f
o
r
eac
h
su
b
-s
a
mp
le.
S
ee E
qu
atio
n
(
8
)
.
T
h
e fi
n
a
n
cial co
n
d
itio
n
e
s
ti
m
ate
d
f
ollo
w
i
n
g
Li et al.
(
2017
)
[
82
]
.
S
ee A
pp
e
n
d
i
x
B
f
o
r
a
m
o
r
e
d
etaile
d
e
x
p
la
n
atio
n
a
n
d
calc
u
late it
w
it
h
E
qu
atio
n
(
A2
)
.
T
h
e co
ns
e
r
vati
sm
s
co
r
e e
s
ti
m
ate
d
f
ollo
w
i
n
g
K
h
a
n
a
n
d
W
att
s
(
2009
)
[
77
]
.
S
ee A
pp
e
n
d
i
x
B
f
o
r
a
m
o
r
e
d
etaile
d
e
x
p
la
n
atio
n
a
n
d
calc
u
late it
w
it
h
E
qu
atio
n
(
A1
)
.
M
e
as
ur
e o
f
s
to
c
k
r
et
ur
n
s
y
nch
r
o
n
i
c
ity
d
evelo
p
e
d
b
y
M
o
r
c
k et
a
l.
(
2000
)
[
86
]
.
S
ee A
pp
e
n
d
ix
B
f
o
r
a
m
o
r
e
d
etaile
d
e
x
p
la
n
atio
n
a
n
d
calc
u
late it
w
it
h
E
qu
atio
ns
(
A3
)
a
n
d
(
A4
)
.
T
h
e
n
at
ur
al lo
g
a
r
it
hm
o
f
t
h
e co
mp
a
n
y
’
s
total a
ss
et
s
f
o
r
eac
h
s
a
mp
le yea
r
.
It
r
e
f
e
rs
to t
h
e book val
u
e o
f
all liabilitie
s
d
ivi
d
e
d
by total a
ss
et
s
at t
h
e e
n
d
o
f
a fi
s
cal yea
r
.
A
r
atio
r
e
pr
e
s
e
n
ti
n
g
t
h
e e
f
ficie
n
cy o
f
t
h
e
us
e o
f
sh
a
r
e
h
ol
d
e
rs’
fun
d
s
.
T
h
e
m
a
r
k
e
t
-
to
-
boo
k
r
atio o
f
fi
rm
i i
n
y
e
a
r
t, t
h
at is,
(m
a
r
k
e
t
pr
ic
e
at t
h
e
e
nd
o
f
fiscal y
e
a
r
×
num
b
e
r
o
f
sh
a
r
e
s
o
u
t
s
ta
n
d
i
n
g
+
n
et a
ss
et val
u
e
p
e
r
sh
a
r
e ×
num
be
r
o
f
n
o
n-
t
r
a
d
able o
u
t
s
ta
n
d
i
n
g
sh
a
r
e
s)
/book
val
u
e o
f
e
qu
ity.
A
m
e
as
ur
e o
f
fi
nanc
i
a
l
r
e
p
o
r
ti
n
g o
p
aq
u
e
n
e
ss
w
h
o
s
e
ca
l
c
u
l
a
tio
n
i
s
t
h
e 3
-
ye
a
r
m
ovi
n
g
s
u
m
o
f
t
h
e
ab
s
ol
u
te val
u
e o
f
a
nnu
al
p
e
rf
o
rm
a
n
ce
-
a
d
j
us
te
d
d
i
s
c
r
etio
n
a
r
y acc
ru
al
s
.
It i
s
a
pr
oxy
f
o
r
t
h
e level o
f
i
n
ve
s
to
r
d
i
sa
g
r
ee
m
e
n
t
w
h
o
s
e
ca
l
c
u
l
a
tio
n
i
s
t
h
e
a
ve
r
a
ge
m
o
n
t
h
ly
sha
r
e
t
ur
n
ove
r
ove
r
t
h
e
c
urr
e
n
t fi
sca
l ye
a
r
m
i
n
u
s
t
h
e
a
ve
r
a
ge
m
o
n
t
h
ly
sha
r
e t
ur
n
ove
r
ove
r
t
h
e
pr
evio
u
s
y
e
a
r
,
wh
e
r
e
m
o
n
t
h
ly s
h
a
r
e
t
urn
ov
e
r
is calc
u
lat
e
d
as t
h
e
m
o
n
t
h
ly s
h
a
r
e
-
t
r
a
d
i
n
g
vol
um
e
d
ivi
d
e
d
by t
h
e
num
be
r
o
f
sh
a
r
e
s
o
u
t
s
ta
n
d
i
n
g
ove
r
t
h
e
m
o
n
t
h
.
T
h
e HFT
m
a
r
ket
s
ize i
n
t
h
e U.
S
.
fr
o
m
t
h
e I
S
I
B
W
o
r
l
d
w
eb
s
ite
;
t
h
e HFT i
s
m
ea
sur
e
d
by o
n
e t
h
o
us
a
n
d
billio
n
d
olla
rs
.
A
pp
e
nd
i
x
B. Co
r
p
o
r
ate Gove
r
n
a
n
ce a
nd
Sy
n
c
h
r
o
n
icity
A
pp
en
d
i
x
B
.1.
M
e
a
su
r
ement
o
f
Fi
r
m
-
S
p
eci
f
i
c Acc
o
unting C
o
nse
rv
a
tism
(
Csc
o
r
e
)
W
e
m
e
as
ur
e t
h
e
d
eg
r
ee o
f
acc
o
u
n
ti
n
g
c
o
ns
e
r
v
a
ti
sm
f
o
r
e
ach
sam
p
le
an
d
e
ach
fi
r
m
u
s
i
n
g t
h
e fi
r
m-
ye
a
r
c
o
n
d
itio
na
l
c
o
ns
e
r
v
a
ti
sm
m
e
as
ur
e C
sc
o
r
e
d
evelo
p
e
d
b
y K
han
an
d
W
att
s
(
2009
)
[
77
]
. Co
n
c
r
e
t
e
ly, to acq
u
i
r
e
t
h
e
C
s
co
r
e
m
e
a
sur
e
,
w
e
obtai
n
t
h
e
co
e
f
fici
e
n
t
s
t
h
at
Ent
r
o
p
y
2025, 27, 345
23 o
f
26
a
r
e
c
oll
a
te
d
w
it
h
c
o
ns
e
r
v
a
ti
sm
b
y e
s
ti
ma
ti
n
g t
h
e
f
ollo
w
i
n
g
r
eg
r
e
ss
io
n
, t
ha
t i
s
, γ
1
, γ
1
, γ
1
,
a
n
d
γ
1
at ti
m
e t
:
i
Y = β
1
+ β
2
× D
i
+ R
i
× (µ
1
+ µ
2
× Si
z
e
i
+ µ
3
×
MB
i
+ µ
4
× Le
v
e
r
a
ge
i
)
+D
i
× R
i
× (γ
1
+ γ
2
× Si
z
e
i
+ γ
3
×
MB
i
+ γ
4
× Le
v
e
r
a
ge
i
)
+(
δ
1
× Si
z
e
i
+
δ
2
×
MB
i
+
δ
3
× Le
v
e
r
a
ge
i
+
δ
4
× D
i
× Si
z
e
i
+
δ
5
× D
i
×
BM
i
+
δ
6
× D
i
× Le
v
e
r
a
ge
i
) +
ε
i
,
(
A1
)
iii
t,5
t,5
wh
e
r
e
Y i
s
m
e
a
sur
e
d
a
s
E
P
S/
P
,t−1
,
w
it
h
P
,t−1
i
s
t
h
e
s
tock
pr
ic
e
at t
h
e
b
e
g
i
nn
i
n
g
o
f
ti
m
e
t,
a
n
d
E
P
S i
s
t
h
e ea
rn
i
n
g
s
p
e
r
sh
a
r
e
m
ea
sur
e
d
by o
p
e
r
ati
n
g
pr
ofit
m
ea
sur
e
d
by t
h
e
num
be
r
o
f
s
tock
s
o
u
t
s
ta
nd
i
n
g
. R
i
i
s
t
h
e
b
u
y
-
a
nd-h
ol
d
yi
e
l
d
, a
d
j
us
t
e
d
f
o
r
t
h
e
co
rr
e
sp
o
nd
i
n
g
m
a
r
k
e
t
r
et
u
rn
,
fr
o
m
t
h
e fi
f
t
h
m
o
n
t
h
a
f
te
r
t
h
e e
n
d
o
f
fi
s
cal yea
r
t to t
h
e
f
o
u
r
t
h
m
o
n
t
h
o
f
yea
r
t + 1, R
i
=
∑
t+1,4
(1 +
r
i,t
)−
∑
t+1,4
(1 +
r
m,t
),
w
h
e
r
e
r
i,t
i
s
t
h
e fi
r
m
m
o
n
t
h
ly
r
et
ur
n
, i
nc
l
ud
i
n
g
d
ivi
d
e
nds
, a
nd
r
m,t
i
s
t
h
e
m
o
n
t
h
ly
m
a
r
ket val
u
e
-
w
ei
g
h
te
d
r
et
urn
. D
i
i
s
a
dumm
y va
r
iable,
s
et to o
n
e i
f
R
i
< 0 a
n
d
to ze
r
o ot
h
e
r
w
i
s
e. Si
z
e i
s
t
h
e
n
at
ur
al lo
g
o
f
t
h
e book val
u
e o
f
total
ass
e
r
ts.
MB
is t
h
e
m
a
r
k
e
t
-
to
-
boo
k
r
atio, a
nd
l
e
v
e
r
a
g
e
is t
h
e
liability
-
to
-
ass
e
t
r
atio. T
h
e
n
,
w
e
o
b
tai
n
a
m
ea
s
u
r
e o
f
fi
rm-
yea
r
c
o
n
d
itio
n
al
c
o
ns
e
r
vati
sm
(
C
sc
o
r
e
)
, a
pp
lyi
n
g
t
h
e
f
ollo
w
i
n
g
li
n
ea
r
fun
ctio
ns
:
Csc
o
r
e
i
= γ
1
+ γ
2
× Si
z
e
i
+ γ
3
×
M
T
B
i
+ γ
4
× Le
v
e
r
a
ge
i
,
(
A2
)
I
f
a fi
rm
h
a
s
a
h
i
g
h
C
s
co
r
e, it i
s
co
ns
i
d
e
r
e
d
m
o
r
e co
ns
e
r
vative.
A
pp
en
d
i
x
B
.2.
M
e
a
su
r
ement
o
f
Fi
r
m
-
S
p
eci
f
i
c Fin
a
nci
a
l Dist
r
ess
(
Dist
r
ess
)
M
a
n
y st
ud
i
e
s
u
s
e
t
h
e
Z
-
sco
r
e
d
e
fi
n
e
d
by Alt
m
a
n
(
1968
)
[
87
]
,
wh
ic
h
is
u
s
u
ally t
r
e
at
e
d
as
t
h
e
m
ost
w
i
d
e
ly
u
s
e
d
fi
n
a
n
cial
d
ist
r
e
ss
m
e
as
ur
e
[
88
]
. Za
n
g
(
2012
)
[
81
]
m
o
d
ifi
e
d
t
h
e
Z
-
sco
r
e
to
pr
oxy
f
o
r
a
fi
r
m’s
fi
nanc
i
a
l
h
e
a
lt
h
. Follo
w
i
n
g Z
an
g
(
2012
)
[
81
]
,
w
e
u
s
e t
h
e
m
o
d
ifie
d
Z
-s
co
r
e to
m
ea
sur
e fi
n
a
n
cial
d
i
s
t
r
e
ss
:
ZSc
o
r
e
i,t
= 0.3X
1
+ 1.0X
2
+ 1.4X
3
+ 1.2X
4
+ 0.6X
5
,
(
A3
)
w
h
e
r
e Dist
r
ess
d
e
n
ote
s
fi
nanc
i
a
l
d
i
s
t
r
e
ss
,
an
d
X
1
i
s
n
et
pr
ofit ove
r
tot
a
l
ass
et
s
. X
2
i
s
sa
le
s
ove
r
tot
a
l
ass
et
s
; X
3
i
s
r
et
a
i
n
e
d
e
a
r
n
i
n
g
s
ove
r
tot
a
l
ass
et
s
; X
4
i
s
w
o
r
ki
n
g
ca
p
it
a
l
ove
r
tot
a
l
ass
et
s
an
d
X
5
i
s
t
h
e
ma
r
ket v
a
l
u
e o
f
e
q
u
ity ove
r
tot
a
l li
ab
ilitie
s
. I
f
ZSc
o
r
e
i,t
i
s
l
a
r
ge
r
, t
h
e fi
nanc
i
a
l
c
o
n
d
itio
n
o
f
fi
r
ms
i
s
b
ette
r
. T
h
e −ZSc
o
r
e
i,t
i
s
little,
an
d
t
h
e fi
r
m
i
s
i
n
fi
n
a
n
cial
d
i
s
t
r
e
ss
.
A
pp
en
d
i
x
B
.3.
M
e
a
su
r
ement
o
f
Fi
r
m
-
S
p
eci
f
i
c St
o
ck
P
r
ice S
y
nc
h
r
o
nicit
y
S
toc
k
sy
n
c
hr
o
n
icity
r
e
fl
e
cts i
nf
o
rm
atio
n
asy
mm
e
t
r
y b
e
t
w
ee
n
co
mp
a
n
i
e
s a
nd
i
n
v
e
sto
r
s.
To
m
ea
sur
e
s
tock
pr
ice
s
y
n
c
hr
o
n
icity,
w
e fi
rs
t calc
u
late t
h
e R
2
o
f
co
mp
a
n
y i i
n
fi
s
cal yea
r
t
by
us
i
n
g
t
h
e
r
e
g
r
e
ss
io
n
o
f
t
h
e
m
a
r
ket
s
i
n
g
le
-f
acto
r
m
o
d
el
[
89
]
.
i
t
RET
,t
=
α
+ β
M
KTRET +
ε
i,t
,
(
A4
)
i
t
2
2
w
h
e
r
e RET
,t
i
s
t
h
e
r
et
urn
r
atio o
f
fi
rm
i i
n
w
eek t, a
n
d
M
KTRET i
s
t
h
e
m
a
r
ket ca
p
italiza
-
tio
n
w
eig
h
te
d
r
et
ur
n
r
a
tio
n
i
n
a
w
eek. T
h
e
d
a
t
a
s
ele
c
te
d
h
e
r
e
a
r
e
ma
i
n
ly
S&P
500 i
n
d
ex
d
a
t
a
w
eig
h
te
d
b
y t
h
e flo
a
ti
n
g
ma
r
ket v
a
l
u
e. R
i,t
f
o
r
e
ach
fi
r
m
eve
r
y ye
a
r
i
s
o
b
t
a
i
n
e
d
b
y
r
eg
r
e
ss
io
n
.
S
i
nc
e t
h
e
r
an
ge o
f
R
i,t
i
s
[
0,1
]
,
f
ollo
w
i
n
g
M
o
r
c
k et
a
l.
(
2000
)
[
86
]
, t
h
i
s
p
a
p
e
r
d
efi
n
e
s
t
h
e
s
tock
pr
ice
s
y
n
c
hr
o
n
izatio
n
i
n
d
e
x
o
f
t
h
e fi
rm
i at ti
m
e t.
2
1 − R
2
S
y
nc
h
i,t
= l
n
(
R
i,t
),
(
A5
)
i,t
Ent
r
o
p
y
2025, 27, 345
24 o
f
26
A
h
i
g
h
S
y
nc
h
i,t
i
n
d
icate
s
t
h
at t
h
e fi
rm
i
s
h
i
g
h
ly co
rr
elate
d
w
it
h
t
h
e
m
a
r
ket.
Refe
r
e
n
ce
s
1. C
h
o
n
g
, E.
;
Ho, C.C.
;
O
n
g
, Z.F.
;
O
n
g
, H.H. U
s
i
n
g
n
e
w
s
s
e
n
ti
m
e
n
t
f
o
r
eco
n
o
m
ic
f
o
r
eca
s
ti
n
g
:
A
M
alay
s
ia
n
ca
s
e
s
t
u
d
y. I
n
M
a
c
h
ine
Le
a
r
ning in Cent
r
a
l
B
a
nking
;
B
a
n
k
f
o
r
I
n
t
e
rn
atio
n
al
S
e
ttl
e
m
e
n
t
s
, E
d
.
;
Vol
um
e
57 o
f
IFC
Bu
ll
e
ti
ns
C
h
a
p
t
e
rs
;
B
a
n
k
f
o
r
I
n
t
e
rn
atio
n
al
S
ettle
m
e
n
t
s
:
B
a
s
el,
S
w
itze
r
la
n
d
, 2022.
2. Co
rr
e
a
, R.
;
G
a
rud
, K.
;
Lo
n
d
o
n
o,
J
.
M
.
;
M
i
s
l
an
g, N.
S
e
n
ti
m
e
n
t i
n
c
e
n
t
r
a
l
ban
k
s’
fi
nanc
i
a
l
s
t
ab
ility
r
e
p
o
r
t
s
. Re
v
. Fi
n
a
n
c. 2021,
25, 85–120.
3.
B
e
nn
a
n
i, H. Ce
n
t
r
al ba
n
k co
mmun
icatio
n
i
n
t
h
e
m
e
d
ia a
n
d
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t. J. Ec
o
n.
B
e
ha
v
.
O
r
g
a
n. 2020, 176, 431–444.
4.Ja
n
g
, J.
;
Ka
n
g
, J.
P
r
obability o
f
pr
ic
e
c
r
a
sh
e
s
,
r
atio
n
al
sp
e
c
u
lativ
e
b
u
bbl
e
s
, a
n
d
t
h
e
c
r
o
ss-s
e
ctio
n
o
f
s
tock
r
e
t
urns
. J. Fin
a
n
c
. E
c
o
n.
2019, 132, 222–247.
[
C
r
o
ss
Re
f]
5.Aba
k
a
h
, E.J.A.
;
Ab
du
lla
h
,
M
.
;
Yo
u
sa
f
, I.
;
Ti
w
a
r
i, A.K.
;
Li, Y. Eco
n
o
m
ic sa
n
c
t
io
n
s
n
e
w
s s
e
n
t
i
m
e
n
t
s a
nd
g
lobal s
t
oc
k
m
a
r
k
et
s
dur
i
n
g
t
h
e R
uss
ia
-
Uk
r
ai
n
e co
n
flict. J. Int. Fin
a
nc.
M
a
r
k. Inst.
M
o
ne
y
2023, 91, 101910.
6.
B
ake
r
,
M
.
;
Wur
g
le
r
, J. I
n
ve
s
to
r
s
e
n
ti
m
e
n
t i
n
t
h
e
s
tock
m
a
r
ket. J. Ec
o
n.
P
e
r
s
p
ect. 2007, 21, 129–151.
[
C
r
o
ss
Re
f]
7.
O
ECD. I
mpac
ts
o
f
t
h
e Russ
i
a
n In
v
a
s
i
o
n
o
f
U
k
r
a
i
ne
o
n
Fi
n
a
n
c
i
a
l
M
a
r
ket C
o
n
d
i
t
i
o
ns
a
n
d
Res
ili
en
c
e
;
T
e
c
hn
ical R
e
p
o
r
t
;
O
ECD
P
u
bli
sh
i
n
g
:
P
a
r
i
s
, F
r
a
n
ce, 2022.
[
C
r
o
ss
Re
f]
8.
B
r
an
d
t,
M
.
W
.
;
G
a
o, L.
Mac
r
o
f
u
n
d
am
e
n
t
a
l
s
o
r
geo
p
oliti
ca
l eve
n
t
s
? A text
u
a
l
ana
ly
s
i
s
o
f
n
e
w
s
eve
n
t
s
f
o
r
c
rud
e oil.
J
. Empi
r
.
Fin
a
nc. 2019, 51, 64–94.
9.
W
a
n
g
, X.
;
Wu
, Y.
;
X
u
,
W
. Geo
p
olitical
r
i
s
k a
n
d
i
n
ve
s
t
m
e
n
t. J.
M
o
ne
y
C
r
e
d
it
B
a
nk. 2024, 56 , 2023–2059.
10.
W
elc
h
, I. T
h
e
w
i
s
d
o
m
o
f
t
h
e
r
obi
nh
oo
d
c
r
o
wd
. J. Fin
a
nc. 2022, 77, 1489–1527.
11.
B
a
r
be
r
i
s
, N.
;
Sh
lei
f
e
r
, A.
;
Vi
shn
y, R. A
m
o
d
el o
f
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t. J. Fin
a
nc. Ec
o
n. 1998, 49, 307–343.
12.C
h
a
n
g
, X.
;
C
h
e
n
, Y.
;
Zolotoy, L.
S
tock li
qu
i
d
ity a
n
d
s
tock
pr
ice c
r
a
sh
r
i
s
k. J. Fin
a
nc.
Q
u
a
nt. An
a
l. 2017, 52, 1605–1637.
13. Reic
hm
a
nn
, D.
;
M
ölle
r
, R.
;
He
r
tel, T. Not
h
i
n
g
b
u
t
g
oo
d
i
n
te
n
tio
ns
:
T
h
e
s
ea
r
c
h
f
o
r
e
qu
ity a
n
d
s
tock
pr
ice c
r
a
sh
r
i
s
k. J.
B
us. Ec
o
n.
2022, 92, 1455–1489.
[
C
r
o
ss
Re
f]
14.
W
e
s
t, K.D. Divi
d
e
n
d
i
nn
ovatio
ns
a
n
d
s
tock
pr
ice volatility. Ec
o
n. J. Ec
o
n. S
o
c. 1988, 56, 37–61.
[
C
r
o
ss
Re
f]
15.
Sh
ille
r
, R.J.
M
a
r
ket volatility a
n
d
i
n
ve
s
to
r
be
h
avio
r
. Am. Ec
o
n. Re
v
. 1990, 80, 58–62.
16.
M
a
x
te
d
,
P
. A
m
ac
r
o
-
fi
n
a
n
ce
m
o
d
el
w
it
h
s
e
n
ti
m
e
n
t. Re
v
. Ec
o
n. Stu
d
. 2023, 91, 438–475.
[
C
r
o
ss
Re
f]
17.K
a
lyv
as
, A.
;
P
a
p
a
ky
r
i
a
ko
u
,
P
.
;
S
a
kk
as
, A.
;
U
r
q
u
ha
r
t, A.
Wha
t
dr
ive
s
b
it
c
oi
n’s
pr
i
c
e
c
r
ash
r
i
s
k? Ec
o
n
. Lett. 2019, 191, 108777.
[
C
r
o
ss
Re
f]
18.C
h
i
u
, C.J.
;
Ha
rr
i
s
, R.D.F.
;
S
toja, E.
;
C
h
i
n
,
M
. Fi
n
a
nc
ial
m
a
r
ket volatility,
m
a
cr
oe
c
o
n
o
m
i
c
f
u
n
d
a
m
e
n
tal
s
a
n
d
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t. J.
B
a
nk. Fin
a
nc. 2018, 92, 130–145.
[
C
r
o
ss
Re
f]
19. Li
an
g,
W
.
S
e
ns
itivity to i
n
ve
s
to
r
s
e
n
ti
m
e
n
t
an
d
s
to
c
k
p
e
r
f
o
r
manc
e o
f
o
p
e
n
ma
r
ket
sha
r
e
r
e
pur
chas
e
s
.
J
.
B
a
n
k. Fi
n
a
n
c. 2016,
71, 75–94.
[
C
r
o
ss
Re
f]
20.
B
ake
r
,
M
.
;
Wur
g
le
r
, J.
;
Y
u
a
n
, Y. Global, local, a
n
d
co
n
ta
g
io
us
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t. J. Fin
a
nc. Ec
o
n. 2012, 104, 272–287.
[
C
r
o
ss
Re
f]
21.H
an
,
M
.
;
Z
h
o
u
,
J
.
M
u
lti
-
S
ca
le C
ha
r
ac
te
r
i
s
ti
cs
o
f
I
n
ve
s
to
r
S
e
n
ti
m
e
n
t T
r
ansm
i
ss
io
n
Bas
e
d
o
n
Wa
velet, T
r
ansf
e
r
E
n
t
r
o
p
y
an
d
Net
w
o
r
k A
n
aly
s
i
s
. Ent
r
o
p
y
2022, 24, 1786.
[
C
r
o
ss
Re
f]
22.
B
illio,
M
.
;
Get
m
a
ns
ky,
M
.
;
Lo, A.
W
.
;
P
elizzo
n
, L. E
c
o
n
o
m
et
r
i
c
m
ea
s
u
r
e
s
o
f
c
o
nn
e
c
te
d
n
e
ss
a
n
d
s
y
s
te
m
i
c
r
i
s
k i
n
t
h
e fi
n
a
nc
e a
n
d
i
nsur
a
n
ce
s
ecto
rs
. J. Fin
a
nc. Ec
o
n. 2012, 104, 535–559.
[
C
r
o
ss
Re
f]
23.Cao, J.
;
W
e
n
, F.
;
E
u
g
e
n
e,
S
.H.
M
ea
sur
i
n
g
t
h
e
s
y
s
te
m
ic
r
i
s
k i
n
i
n
d
i
r
ect fi
n
a
n
cial
n
et
w
o
r
k
s
. Eu
r
. J. Fin
a
nc. 2022, 28, 1053–1098.
24.El
-
K
h
a
t
ib, R.
;
F
o
g
e
l, K.
;
Ja
nd
i
k
, T. CE
O
n
et
w
o
r
k
c
e
n
t
r
ali
t
y a
nd
m
e
r
g
e
r
p
e
rf
o
rm
a
n
c
e
. J.
Fi
n
a
n
c
. E
c
o
n. 2015, 116, 349–382.
[
C
r
o
ss
R
e
f]
25.
B
altaky
s
, K.
;
Le Viet, H.
;
Ka
nn
iai
n
e
n
, J.
S
t
ru
ct
ur
e o
f
I
n
ve
s
to
r
Net
w
o
r
k
s
a
n
d
Fi
n
a
n
cial C
r
i
s
e
s
. Ent
r
o
p
y
2021, 23, 381.
[
C
r
o
ss
Re
f]
26.
G
ao, X.
;
H
u
a
n
g
,
S
.
;
Sun
, X.
;
Hao, X.
;
A
n
,
F
.
M
o
d
e
lli
n
g
coi
n
te
g
r
a
t
io
n
a
nd
G
r
a
n
g
e
r
ca
u
sali
t
y
n
et
w
o
r
k
t
o
d
ete
c
t
lo
n
g
-
te
rm
e
q
u
ilib
r
i
um
a
n
d
d
i
ffus
io
n
p
at
hs
i
n
t
h
e fi
n
a
n
cial
s
y
s
te
m
. R. S
o
c.
O
p
en Sci. 2018, 5, 172092.
[
P
u
b
M
e
d
]
27.C
h
e
n
,
M
.
;
W
a
n
g
, Y.
;
Wu
,
B
.
;
H
u
a
n
g
, D. Dy
n
a
m
ic A
n
aly
s
e
s
o
f
Co
n
t
a
g
io
n
Ri
s
k a
nd
M
o
du
l
e
Evol
u
t
io
n
o
n
t
h
e
SS
E A
-Sh
a
r
e
s
M
a
r
k
et
B
a
s
e
d
o
n
M
i
n
i
mum
I
nf
o
rm
atio
n
E
n
t
r
o
p
y. Ent
r
o
p
y
2021, 23, 434.
[
C
r
o
ss
Re
f]
28. Go
n
g, X.
-
L.
;
Li
u
,
J
.
-M
.
;
Xio
n
g, X.
;
Z
han
g,
W
. Re
s
e
a
r
ch
o
n
s
to
c
k vol
a
tility
r
i
s
k
an
d
i
n
ve
s
to
r
s
e
n
ti
m
e
n
t
c
o
n
t
a
gio
n
f
r
o
m
t
h
e
p
e
rsp
ective o
f
mu
lti
-
laye
r
d
y
n
a
m
ic
n
et
w
o
r
k. Int. Re
v
. Fin
a
nc. An
a
l. 2022, 84, 102359.
29.Jack
s
o
n
,
M
.
O
. S
o
ci
a
l
a
n
d
Ec
o
n
o
mic Netw
o
r
ks
;
P
r
i
n
ceto
n
U
n
ive
rs
ity
P
r
e
ss
:
P
r
i
n
ceto
n
, NJ, U
S
A, 2008
;
Vol
um
e 3.
30.
P
itt
m
a
n
, J.
;
W
a
n
g
, L.
;
Wu
, D. Net
w
o
r
k a
n
aly
s
i
s
o
f
a
u
d
it
p
a
r
t
n
e
r
r
otatio
n
. C
o
ntem
p
. Acc
o
unt. Res. 2022, 39, 1085–1119.
31.
Janc
ze
w
s
ki, A.
;
A
na
g
n
o
s
to
u
, I.
;
K
an
d
ha
i, D. I
nf
e
rr
i
n
g De
a
le
r
Net
w
o
r
k
s
i
n
t
h
e Fo
r
eig
n
Ex
chan
ge
Ma
r
ket U
s
i
n
g Co
n
d
itio
na
l
T
r
a
nsf
e
r
E
n
t
r
o
p
y
:
A
n
aly
s
i
s
o
f
a Ce
n
t
r
al
B
a
n
k A
nn
o
un
ce
m
e
n
t. Ent
r
o
p
y
2024, 26, 738.
[
C
r
o
ss
Re
f]
32.Goe
r
g
e
n
,
M
.
;
Re
nn
eboo
g
, L.
;
Z
h
ao, Y. I
ns
i
d
e
r
t
r
a
d
i
n
g
a
n
d
n
et
w
o
r
ke
d
d
i
r
ecto
rs
. J. C
o
r
p
. Fin
a
nc. 2019, 56, 152–175.
33. K
ama
te, V.
;
K
u
ma
r
, A. De
a
le
r
n
et
w
o
r
k
s
,
c
lie
n
t
s
o
p
h
i
s
ti
ca
tio
n
an
d
pr
i
c
i
n
g i
n
OTC
d
e
r
iv
a
tive
s
.
J
. I
n
t.
M
o
n
e
y
Fi
n
a
n
c. 2024,
140, 102986.
Ent
r
o
p
y
2025, 27, 345
25 o
f
26
34. G
a
o,
Q
.
;
F
an
, H.
;
Y
u
, C.
S
y
s
te
m
i
c
I
m
p
o
r
t
anc
e
an
d
Ri
s
k C
ha
r
ac
te
r
i
s
ti
cs
o
f
Ban
k
s
Bas
e
d
o
n
a
M
u
lti
-
L
a
ye
r
Fi
nanc
i
a
l Net
w
o
r
k
A
n
aly
s
i
s
. Ent
r
o
p
y
2024, 26, 378.
[
C
r
o
ss
Re
f]
[
P
u
b
M
e
d
]
35.Go
d
i
g
be,
B
.G.
;
C
hu
i, C.
M
.
;
Li
u
, C.L. Di
r
ecto
rs
n
et
w
o
r
k ce
n
t
r
ality a
n
d
ea
rn
i
n
g
s
qu
ality. A
pp
l. Ec
o
n. 2018, 50, 5381–5400.
36.Co
s
te
n
ba
d
e
r
, E.
;
Vale
n
te, T.
W
. T
h
e
s
tability o
f
ce
n
t
r
ality
m
ea
sur
e
s
w
h
e
n
n
et
w
o
r
k
s
a
r
e
s
a
mp
le
d
. S
o
c. Netw. 2003, 25, 283–307. 37.
A
u
d
r
ito,
G
.
;
P
ia
n
i
n
i, D.
;
Da
m
ia
n
i,
F
.
;
Vi
r
oli,
M
. A
gg
r
e
g
at
e
c
e
n
t
r
ality
m
e
a
sur
e
s
f
o
r
IoT
-
ba
s
e
d
coo
r
d
i
n
atio
n
.
S
c
i. C
o
m
p
ut.
P
r
o
g
r
a
m.
2021, 203, 102584.
38. Ya
n
, X.L.
;
C
u
i, Y.
P
.
;
Ni,
S
.J. I
d
e
n
ti
f
yi
n
g
i
n
fl
u
e
n
tial
spr
ea
d
e
rs
i
n
co
mp
le
x
n
et
w
o
r
k
s
ba
s
e
d
o
n
e
n
t
r
o
p
y
w
ei
g
h
t
m
et
h
o
d
a
n
d
g
r
avity
la
w
. C
h
in.
P
h
y
s.
B
2020, 29, 048902.
39.H
w
a
n
g
, C.L.
;
Yoo
n
, K.
M
ulti
p
le Att
r
i
b
ute Decisi
o
n
M
a
king
:
M
et
h
o
d
s
a
n
d
A
pp
lic
a
ti
o
ns
;
Spr
i
n
g
e
r
:
Ne
w
Y
o
r
k, NY, U
S
A, 1981.
40. Yo
us
a
f
, I.
;
Yo
uss
e
f
,
M
.
;
Goo
d
ell, J.
W
.
Q
u
a
n
tile co
nn
ecte
d
n
e
ss
bet
w
ee
n
s
e
n
ti
m
e
n
t a
n
d
fi
n
a
n
cial
m
a
r
ket
s
:
Evi
d
e
n
ce
fr
o
m
t
h
e
S&
P
500 t
w
itte
r
s
e
n
ti
m
e
n
t i
n
d
e
x
. Int. Re
v
. Fin
a
nc. An
a
l. 2022, 83, 102322.
41. Da
r
i
n
g
, J.R. C
r
isis
m
a
n
a
g
e
m
e
n
t
i
n
i
n
te
rn
a
t
io
n
al b
u
si
n
e
ss
:
K
e
ys
t
o
e
ff
e
c
t
iv
e
d
e
cisio
n
m
a
k
i
n
g
. Le
ad
e
r
s
h
.
Or
g
a
n. De
v
. J. 1994, 15, 3–8.
42. X
u
, N.
;
Li, X.
;
Y
u
an
,
Q
.
;
C
han
, K.C. Ex
c
e
ss
p
e
r
k
s
an
d
s
to
c
k
pr
i
c
e
c
r
ash
r
i
s
k
:
Evi
d
e
nc
e
f
r
o
m
C
h
i
na
.
J
. C
o
r
p. Fi
n
a
n
c. 2014,
25, 419–434.
43.A
n
, H.
;
Z
h
a
n
g
, T.
S
tock
pr
ice
s
y
n
c
hr
o
n
icity, c
r
a
sh
r
i
s
k, a
n
d
i
ns
tit
u
tio
n
al i
n
ve
s
to
rs
. J. C
o
r
p
. Fin
a
nc. 2013, 21, 1–15.
44. D
ur
n
ev, A.
;
M
o
r
c
k, R.
;
Ye
u
n
g,
B
.
;
Z
a
r
o
w
i
n
,
P
. Doe
s
g
r
e
a
te
r
fi
r
m-s
p
e
c
ifi
c
r
et
ur
n
v
a
r
i
a
tio
n
m
e
an
m
o
r
e o
r
le
ss
i
nf
o
r
m
e
d
s
to
c
k
pr
ici
n
g
?
J. Acc
o
unt. Res. 2003, 41, 797–836.
45.K
um
a
r
, A.
;
Lee, C.
M
.C. Retail i
n
ve
s
to
r
s
e
n
ti
m
e
n
t a
n
d
r
et
urn
co
m
ove
m
e
n
t
s
. J. Fin
a
nc. 2006, 61, 2451–2486.
[
C
r
o
ss
Re
f]
46.De, L.J.
B
.
;
Sh
lei
f
e
r
, A.
;
Summ
e
rs
, L.H. Noi
s
e t
r
a
d
e
r
r
i
s
k i
n
fi
n
a
n
cial
m
a
r
ket
s
. J.
P
o
litic
a
l Ec
o
n. 1990, 98, 703–738.
47.Ha
n
,
B
. I
n
ve
s
to
r
s
e
n
ti
m
e
n
t a
n
d
o
p
tio
n
pr
ice
s
. Re
v
. Fin
a
nc. Stu
d
. 2008, 21, 387–414.
[
C
r
o
ss
Re
f]
48.
P
h
a
m
,
M
.D.
M
.
;
N
gu
ye
n
, T.H.
Wh
e
n
a
g
i
f
t
r
e
s
e
m
ble
s
a t
r
oja
n
h
o
rs
e
:
CE
O
s
tock
g
i
f
t a
n
d
s
tock
p
r
ice c
r
a
sh
r
i
s
k.
Br
. Acc
o
un
t. Re
v
.
2023, 56, 101235.
[
C
r
o
ss
Re
f]
49.Li, D.
;
Xi
n
g, L.
;
Z
ha
o, Y. Doe
s
exte
n
d
e
d
a
ud
ito
r
d
i
sc
lo
s
ur
e
d
ete
r
mana
ge
r
i
a
l
ba
d
-n
e
w
s
h
o
a
rd
i
n
g? Evi
d
e
nc
e
f
r
o
m
c
r
ash
r
i
s
k. J.
C
o
r
p
. Fin
a
nc. 2022, 76, 102256.
[
C
r
o
ss
Re
f]
50.
B
o
u
bak
e
r
,
S
.
;
M
a
ns
ali, H.
;
Rjiba, H. La
r
g
e
co
n
t
r
olli
n
g
sh
a
r
e
h
ol
d
e
rs
a
nd
s
tock
pr
ic
e
s
y
n
c
hr
o
n
icity. J.
B
a
nk.
Fi
n
a
n
c
. 2014, 40, 80–96.
[
C
r
o
ss
Re
f]
51.
B
ake
r
,
M
.
;
S
tei
n
, J.C.
M
a
r
ket li
qu
i
d
ity a
s
a
s
e
n
ti
m
e
n
t i
n
d
icato
r
. J. Fin
a
nc.
M
a
r
k. 2004, 7, 271–299.
[
C
r
o
ss
Re
f]
52.
P
e
n
g
, L.
;
Z
h
a
n
g
, L.
U
n
l
e
a
sh
i
n
g
t
h
e
c
r
o
w
d
:
T
h
e
e
ff
e
ct o
f
s
ocial
n
e
t
w
o
r
k
s
i
n
c
r
o
w
dfund
i
n
g
m
a
r
k
e
t
s
.
M
a
n
a
g.
S
c
i
. 2024..
[
C
r
o
ss
R
e
f]
53.D
u
a
n
, J.
;
Li
n
, J. T
h
e
i
mp
ac
t
o
f
C
O
VID
-
19 o
n
t
h
e
c
r
as
h
r
is
k
o
f
r
e
g
is
te
r
e
d
n
e
w
s
h
a
r
e
s i
n
C
h
i
n
a.
P
ac
.
B
a
s
i
n
Fi
n
a
n
c
. J. 2023, 79, 102037.
[
C
r
o
ss
Re
f]
54.Z
h
a
n
g
,
P
.
;
Yi
n
,
S
.
;
Sh
a, Y. Global
s
y
s
t
e
m
ic
r
i
s
k
d
y
n
a
m
ic
n
e
t
w
o
r
k co
nn
e
ct
e
d
n
e
ss
d
ur
i
n
g
t
h
e
C
O
VID
-
19
:
Evi
d
e
n
c
e
fr
o
m
n
o
n
li
n
e
a
r
G
r
a
n
g
e
r
ca
us
ality. J. Int. Fin
a
nc.
M
a
r
k. Inst.
M
o
ne
y
2023, 85, 101783.
[
C
r
o
ss
Re
f]
55.
Br
oock,
W
.A.
;
S
c
h
ei
n
k
m
a
n
, J.A.
;
Dec
h
e
r
t,
W
.D.
;
Le
B
a
r
o
n
,
B
. A te
s
t
f
o
r
i
n
d
e
p
e
n
d
e
n
ce ba
s
e
d
o
n
t
h
e co
rr
elatio
n
d
i
m
e
ns
io
n
. Ec
o
n
.
Re
v
. 1996, 15, 197–235.
[
C
r
o
ss
Re
f]
56.Di
k
s, C.
;
P
a
n
c
h
e
n
k
o, V. A
n
e
w
s
t
a
t
is
t
ic a
nd
pr
ac
t
ical
g
u
i
d
e
li
n
e
s
f
o
r
n
o
np
a
r
a
m
et
r
ic
G
r
a
n
g
e
r
ca
u
sali
t
y
te
s
t
i
n
g
. J. E
c
o
n. D
y
n. C
o
nt
r
o
l
2006, 30, 1647–1669.
57.Y
u
, J.
W
.
;
Xi
e
,
W
.J.
;
Jia
n
g
, Z.
Q
. Ea
r
ly
w
a
rn
i
n
g
m
o
d
e
l ba
s
e
d
o
n
co
rr
e
la
te
d
n
et
w
o
r
k
s
i
n
g
lobal c
rud
e
oil
m
a
r
k
et
s
.
P
h
y
s. A
S
t
a
t.
M
e
ch
.
Its A
pp
l. 2018, 490, 1335–1343.
58.
B
o
n
acic
h
,
P
.
S
o
m
e
un
i
qu
e
pr
o
p
e
r
tie
s
o
f
ei
g
e
n
vecto
r
ce
n
t
r
ality. S
o
c. Netw. 2007, 29, 555–564.
59.C
a
lle
n
,
J
.L.
;
F
an
g, X. I
ns
tit
u
tio
na
l i
n
ve
s
to
r
s
t
ab
ility
an
d
c
r
ash
r
i
s
k
:
M
o
n
ito
r
i
n
g ve
r
s
u
s
sh
o
r
t
-
te
r
m
i
sm
?
J
.
B
a
n
k. Fi
n
a
n
c. 2013,
37, 3047–3063.
60.Ki
m
, Y.
;
Li, H.
;
Li,
S
. Co
rp
o
r
ate
s
ocial
r
e
sp
o
ns
ibility a
n
d
s
tock
pr
ice c
r
a
sh
r
i
s
k. J.
B
a
nk. Fin
a
nc. 2014, 43, 1–13.
61.H
u
y
nh
, T.L.D.
;
F
o
g
lia,
M
.
;
Nasi
r
,
M
.A.
;
A
n
g
e
li
n
i, E.
F
e
v
e
r
is
h
s
e
n
t
i
m
e
n
t
a
nd
g
lobal
e
q
u
i
t
y
m
a
r
k
et
s
dur
i
n
g
t
h
e
C
O
VID
-
19
p
a
nd
e
m
ic. J.
Ec
o
n.
B
e
ha
v
.
O
r
g
a
n. 2021, 188, 1088–1108.
[
C
r
o
ss
Re
f]
62.
W
a
n
g
,
W
.
;
Su
, C.
;
D
u
x
b
ur
y, D. T
h
e
co
nd
i
t
io
n
al i
mp
ac
t
o
f
i
n
v
e
s
t
o
r
s
e
n
t
i
m
e
n
t
i
n
g
lobal
s
t
ock
m
a
r
k
et
s
:
A
t
w
o
-
c
h
a
nn
e
l
e
x
a
m
i
n
a
t
io
n
. J.
B
a
nk. Fin
a
nc. 2022, 138, 106458.
63.Ca
r
d
ellic
h
io,
P
.A. E
s
ti
m
atio
n
o
f
pr
o
d
u
ctio
n
be
h
avio
r
us
i
n
g
p
oole
d
m
ic
r
o
d
ata. Re
v
. Ec
o
n. St
a
t. 1990, 72, 11–18.
64.
Ow
usu-
Gya
p
o
n
g
, A. Alte
rn
ative e
s
ti
m
ati
n
g
tec
hn
i
qu
e
s
f
o
r
p
a
n
el
d
ata o
n
s
t
r
ike activity. Re
v
. Ec
o
n. St
a
t. 1986, 228, 526–531.
65.
B
e
r
g
m
a
n
, N.K.
;
Royc
h
o
wd
hur
y,
S
. I
n
v
e
s
to
r
s
e
n
ti
m
e
n
t a
n
d
co
rp
o
r
at
e
d
i
s
clo
sur
e
. J. A
cc
o
unt. Res. 2008, 46, 1057–1083.
[
C
r
o
ss
Re
f]
66. H
a
o,
J
.
;
Xio
n
g, X. Ret
a
il i
n
ve
s
to
r
a
tte
n
tio
n
an
d
fi
r
ms’
i
d
io
s
y
nc
r
a
ti
c
r
i
s
k
:
Evi
d
e
nc
e
f
r
o
m
C
h
i
na
. I
n
t. Re
v
. Fi
n
a
n
c. A
n
a
l. 2021,
74, 101675.
67.
Bus
ta
m
a
n
te,
M
.C.
;
F
r
é
s
a
r
d
, L. Doe
s
fi
rm
i
n
ve
s
t
m
e
n
t
r
e
sp
o
n
d
to
p
ee
rs’
i
n
ve
s
t
m
e
n
t
?
M
a
n
a
g. Sci. 2021, 67, 4703–4724.
68.C
h
e
n
, Y.
-W
.
;
C
h
a
n
, K.
;
C
h
a
n
g
, Y.
P
ee
r
e
ff
ect
s
o
n
co
rp
o
r
ate ca
sh
h
ol
d
i
n
g
s
. Int. Re
v
. Ec
o
n. Fin
a
nc. 2019, 61, 213–227.
69.
Mu
g
e
rm
a
n
;
Y
e
v
g
e
n
y
;
O
r
ly,
S
.
;
M
o
s
e
s
,
S
. Lo
n
g
t
e
rm
s
avi
n
g
s
d
e
ci
s
io
ns
:
F
i
n
a
n
cial
r
e
f
o
rm
,
p
ee
r
e
ff
e
ct
s
a
n
d
e
t
hn
icity. J. E
c
o
n.
B
e
ha
v
.
O
r
g
a
n. 2014, 106, 235–253.
[
C
r
o
ss
Re
f]
Ent
r
o
p
y
2025, 27, 345
26 o
f
26
70.Ka
us
tia,
M
.
;
Ra
n
tala, V.
S
ocial lea
rn
i
n
g
a
n
d
co
rp
o
r
ate
p
ee
r
e
ff
ect
s
. J. Fin
a
nc. Ec
o
n. 2015, 117, 653–669.
71.
B
e
n
, A.I.
;
S
la
h
e
dd
i
n
e, H. Hi
g
h-fr
e
qu
e
n
cy t
r
a
d
i
n
g
,
s
tock volatility, a
n
d
i
n
t
r
a
d
ay c
r
a
sh
e
s
.
Q
. Re
v
. Ec
o
n. Fin
a
nc. 2022, 84, 337–344.
72. C
a
i, Y.
;
T
an
g, Z.
;
C
h
e
n
, Y. C
an
r
e
a
l
-
ti
m
e i
n
ve
s
to
r
s
e
n
ti
m
e
n
t
h
el
p
pr
e
d
i
c
t t
h
e
h
ig
h-f
r
e
q
u
e
nc
y
s
to
c
k
r
et
ur
ns
? Evi
d
e
nc
e
f
r
o
m
a
m
i
x
e
d
-fr
e
qu
e
n
cy
-r
olli
n
g
d
eco
mp
o
s
itio
n
f
o
r
eca
s
ti
n
g
m
et
h
o
d
. N. Am. J. Ec
o
n. Fin
a
nc. 2024, 72, 102147.
73. Z
h
a
n
g
, F. Hi
g
h-
F
r
e
qu
e
n
cy T
r
a
d
i
n
g
,
S
tock Volatility, a
n
d
P
r
ice Di
s
cove
r
y. 2010. Available o
n
li
n
e
:
h
tt
ps
:
//
p
a
p
e
rs
.
ssrn
.co
m
/
s
ol3
/
p
a
p
e
rs
.c
fm
?ab
s
t
r
act_i
d
=1691679
(
acce
ss
e
d
o
n
10 Feb
ru
a
r
y 2025
)
.
74.
Ja
rr
o
w
, A.R.
;
Pr
otte
r
,
P
. A
d
y
sf
u
nc
tio
na
l
r
ole o
f
h
ig
h
f
r
e
q
u
e
nc
y t
r
a
d
i
n
g i
n
ele
c
t
r
o
n
i
c
ma
r
ket
s
. I
n
t.
J
. The
o
r
. Appl. Fi
n
a
n
c. 2012,
15, 1250022.
[
C
r
o
ss
Re
f]
75.
P
alo
m
i
n
o, F.
;
Re
nn
e
b
oo
g
, L.
;
Z
h
a
n
g
, C. I
nf
o
r
m
atio
n
s
alie
nc
e, i
n
ve
s
to
r
s
e
n
ti
m
e
n
t, a
n
d
s
to
c
k
r
et
ur
ns
:
T
h
e
c
a
s
e o
f
B
r
iti
sh
s
o
cc
e
r
betti
n
g
. J. C
o
r
p
. Fin
a
nc. 2009, 185, 368–387.
76.
M
at
h
ij
s
, C.
;
Rik, F.
S
alie
n
ce t
h
eo
r
y a
n
d
s
tock
pr
ice
s
:
E
mp
i
r
ical evi
d
e
n
ce. J. Fin
a
nc. Ec
o
n. 2021, 140, 460–483.
77.K
h
a
n
,
M
.
;
W
att
s
, R.L. E
s
ti
m
atio
n
a
n
d
e
mp
i
r
ical
pr
o
p
e
r
tie
s
o
f
a fi
rm-
yea
r
m
ea
sur
e o
f
acco
un
ti
n
g
co
ns
e
r
vati
sm
. J. Acc
o
unt. Ec
o
n.
2009, 48, 132–150.
[
C
r
o
ss
Re
f]
78.Ki
m
, J.
B
.
;
W
a
n
g
, Z.
;
Z
h
a
n
g
, L. CE
O
ove
r
co
n
fi
d
e
n
ce a
n
d
s
tock
pr
ice c
r
a
sh
r
i
s
k. C
o
ntem
p
. Acc
o
unt. Res. 2016, 33, 1720–1749.
79.
Sh
lei
f
e
r
, A.
;
Vi
shn
y, R.
W
. La
r
g
e
sh
a
r
e
h
ol
d
e
rs
a
n
d
co
rp
o
r
ate co
n
t
r
ol. J.
P
o
litic
a
l Ec
o
n. 1986, 94
P
t 1, 461–488.
80.Li, Y.
;
Li, X.
;
Xia
n
g
, E.
;
Djaja
d
ike
r
ta, H.G. Fi
n
a
n
cial
d
i
s
t
r
e
ss
, i
n
te
rn
al co
n
t
r
ol, a
n
d
ea
rn
i
n
g
s
m
a
n
a
g
e
m
e
n
t
:
Evi
d
e
n
ce
fr
o
m
C
h
i
n
a. J.
C
o
ntem
p
. Acc
o
unt. Ec
o
n. 2020, 16, 100210.
[
C
r
o
ss
Re
f]
81.Za
n
g
, A.Y. Evi
d
e
n
c
e
o
n
t
h
e
t
r
a
d
e
-
o
ff
b
et
w
ee
n
r
e
al ac
t
ivi
t
i
e
s
m
a
n
i
pu
la
t
io
n
a
nd
acc
ru
al
-
bas
e
d
e
a
rn
i
n
g
s
m
a
n
a
g
e
m
e
n
t
. A
cc
o
unt. Re
v
.
2012, 87, 675–703.
[
C
r
o
ss
Re
f]
82.Li, X.
;
W
a
n
g
,
S
.
S
.
;
W
a
n
g
, X. T
rus
t a
n
d
s
tock
pr
ice c
r
a
sh
r
i
s
k
:
Evi
d
e
n
ce
fr
o
m
C
h
i
n
a. J.
B
a
nk. Fin
a
nc. 2017, 76, 74–91.
83.E
d
m
a
ns
, A.
;
Ga
r
cia, D.
;
No
r
li, Ø.
Sp
o
r
t
s
s
e
n
ti
m
e
n
t a
n
d
s
tock
r
et
urns
. J. Fin
a
nc. 2007, 62, 1967–1998.
84. A
b
ud
y,
M
.
M
.
;
M
u
ge
r
man
, Y.
;
S
h
u
s
t, E. T
h
e
w
i
nn
e
r
T
a
ke
s
it
a
ll
:
I
n
ve
s
to
r
s
e
n
ti
m
e
n
t
an
d
t
h
e E
ur
ovi
s
io
n
S
o
n
g Co
n
te
s
t.
J
.
B
a
n
k.
Fin
a
nc. 2022, 137, 106432.
85.Cook
s
o
n
, J.A.
;
L
u
, R.
;
Mu
lli
ns
,
W
.
;
Nie
ssn
e
r
,
M
. T
h
e
s
ocial
s
i
g
n
al. J. Fin
a
nc. Ec
o
n. 2024, 158, 103870.
[
C
r
o
ss
Re
f]
86.
M
o
r
ck, R.
;
Y
e
un
g
,
B
.
;
Y
u
,
W
. T
h
e
i
nf
o
rm
atio
n
co
n
t
e
n
t o
f
s
tock
m
a
r
k
e
t
s
:
Wh
y
d
o
e
m
e
r
g
i
n
g
m
a
r
k
e
t
s
h
av
e
s
y
n
c
hr
o
n
o
us
s
tock
pr
ic
e
m
ove
m
e
n
t
s
?
J. Fin
a
nc. Ec
o
n. 2000, 58, 215–260.
87.Alt
m
a
n
, E.I. Fi
n
a
n
cial
r
atio
s
,
d
i
s
c
r
i
m
i
n
a
n
t a
n
aly
s
i
s
a
n
d
t
h
e
pr
e
d
ictio
n
o
f
co
rp
o
r
ate ba
n
k
rup
tcy. J. Fin
a
nc. 1968, 23, 589–609.
88.
Bu
g
e
ja,
M
.
;
Cz
e
rn
ko
w
s
ki, R.
;
M
o
r
a
n
, D. T
h
e
i
mp
act o
f
t
h
e
m
a
n
a
g
e
m
e
n
t a
ppr
oac
h
o
n
s
e
g
m
e
n
t
r
e
p
o
r
ti
n
g
. J.
B
us. Fin
a
n
c
. A
cc
o
unt.
2015, 42, 310–366.
89.C
h
a
n
, K.
;
Ha
m
ee
d
, A.
S
tock
pr
ice
s
y
n
c
hr
o
n
icity a
n
d
a
n
aly
s
t cove
r
a
g
e i
n
e
m
e
r
g
i
n
g
m
a
r
ket
s
. J. Fin
a
nc. Ec
o
n. 2006, 80, 115–147.
Di
s
c
l
a
i
m
e
r
/P
ub
li
sh
e
r’
s
N
o
te
:
T
h
e
s
t
a
te
m
e
n
t
s
, o
p
i
n
io
ns
an
d
d
a
t
a
c
o
n
t
a
i
n
e
d
i
n
a
ll
pu
b
li
ca
tio
ns
a
r
e
s
olely t
h
o
s
e o
f
t
h
e i
n
d
ivi
du
a
l
a
u
t
h
o
r(s)
a
nd
co
n
t
r
ib
u
t
o
r(s)
a
nd
n
o
t
o
f
M
D
P
I a
nd
/o
r
t
h
e
e
d
i
t
o
r(s)
.
M
D
P
I a
nd
/o
r
t
h
e
e
d
i
t
o
r(s)
d
i
s
clai
m
r
e
sp
o
ns
ibili
t
y
f
o
r
a
n
y i
n
j
ur
y
t
o
p
eo
p
le o
r
pr
o
p
e
r
ty
r
e
su
lti
n
g
fr
o
m
a
n
y i
d
ea
s
,
m
et
h
o
d
s
, i
ns
t
ru
ctio
ns
o
r
pr
o
d
u
ct
s
r
e
f
e
rr
e
d
to i
n
t
h
e co
n
te
n
t.